Conditions of Identity: A Study in Identity and Survival 0198249748, 9780198249740

Addressing many topics in epistemology and metaphysics, this treatise sets out a new theory of the unity of objects, and

201 71 12MB

English Pages 384 Year 1988

Report DMCA / Copyright


Polecaj historie

Conditions of Identity: A Study in Identity and Survival
 0198249748, 9780198249740

Citation preview

Conditions of Identity A Study in Identity and Survival



Oxford University Press, Walton Street, Oxford OX2 6DP Oxford New York Toronto Delhi Bombay Calcutta Madras Karachi Petaling Jaya Singapore Hong Kong Tokyo Nairobi Dar es Salaam Cape Town Melbourne Auckland and associated companies in Beirut Berlin Ibadan Nicosia Oxford is a trade miJrk of Oxford University Press Published in the United States by Oxford University Press, New York

© Andrew Brennan 1988 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Oxford University Press British Library Cataloguing in Publication Data Brennan, Andrew Conditions of identity: a study in identity and survival. 1. Identity I. Title 111 BD236 ISBN 0-19-824974-8 Library of Congress Cataloging in Publication Data Brennan, Andrew. Conditions of identity. Bibliography: p. Includes index. 1. Identify. 2. Metaphysics. 3. Identify (Psychology) I. Title. BC199.I4B74 1988 110 87-28237 ISBN 0-19-824974-8

Set by Cotswold Typesetting Ltd, Gloucester Printed in Great Britain by Biddies Ltd, Guildford and King's Lynn

Preface PHILOSOPHY is a communal actIvIty in more than one way. Although it can claim to be one of our species' oldest intellectual preoccupations, it cannot advance far without the existence of a global community of philosophers who support, stimulate, and encourage research in the subject. That kind of community itself has links with larger and smaller communities in turn. On the one side there are the smaller, local communities of philosophers supported by their societies in one way or another. On the other side society itself supports philosophy and other intellectual activities because of the value that the study of philosophy, and its practice, is deemed to bring to the community at large. Conscious of this, I have tried to write this book in such a way that, while addressing current concerns in the theory of identity, it is not so technical as to be inaccessible to the non-specialist reader. To some extent, the pursuit of clarity is at odds with the pursuit of accuracy and precision, as many others have noted before me. When faced with a choice, however, I have tried, unashamedly, to err on the side of clarity. I have benefited in preparing this book from the support and assistance of both my local and the wider philosophical community. Over the years at Stirling, I have been fortunate to enjoy the stimulation and encouragement of lively colleagues and friends. They have made contributions to the book in many ways, in some cases long before its preparation was undertaken. As I write this preface, I am acutely aware of the many debts I owe, and of the impossibility of giving them due acknowledgement. The wider community of philosophers also contains many people who have helped me understand both my own, and other people's, views better. Particular gratitude is owed to those colleagues and friends who have gone through the entire manuscript in its draft version, offering suggestions and help without which the book would be a poorer one. Although it is impossible to mention everyone who has helped in this way, I would like to record my special thanks to Jonathan Lowe, Murray MacBeath, AIan Millar, Harold Noonan,



and George Schlesinger, each of whom has made a significant impact on the final version of the work. Thanks also to Elspeth Gillespie for her efficient typing and revision of the manuscript. There are several passages in the book that are based on work that has already appeared in print. I am grateful for permission to reproduce material from the following of my articles: 'Survival', in Synthese, 59 (D. Reidel, 1984); 'Amnesia and Psychological Continuity', in R. Copp and J. J. McIntosh (eds.), New Essays in the Philosophy of Mind (Canadian Journal of Philosophy, Supplementary Volume 11, 1985); 'Best Candidates and Theories of Identity', in Inquiry, 29 (Norwegian University Press, 1986); 'Discontinuity and Identity', in Nous, 21 (Indiana University Press, 1987). I am also grateful to MIT Press for permission to use the epigraph from Word and Object and to Russell Hoban and Pan Books for permission to quote Eusa 33 from Riddley Walker.


Contents Introduction


l. IDENTITY AND CONTINUITY 1.1. Identity and Similarity 1.2. Identity and Continuants 1.3. Continuity and Indiscemibility 1.4. Identity and Survival 1.5. Indeterminancy 1.6. Conditions of Survival

5 5 10 13 17 22 23



2.1. 2.2. 2.3. 2.4. 2.5. 2.6.

The Chain Theory Continuity Problems with Spatio-temporal Continuity Kripke's Disc The Primitiveness of Identity Continuity, Experience, and Unity

3. STRUCTURE AND SURVIVAL 3.1. 3.2. 3.3. 3.4. 3.5. 3.6. 3.7.

The Eusa Problem Telec10ning The Importance of Structure Types and Tokens Replication Production and Copying Processes Pseudomorphism

4. IDENTITY AND DISCONTINUITY 4.1. 4.2. 4.3. 4.4. 4.5. 4.6. 4.7. 4.8. 4.9. 4.10.

Preview Three- and Four-Dimensional Views of the World Episodic Objects and Continuity Scattered Objects and Stages Causation and the Humean Minds, Radio Serials, and Doors A Non-Mereological Solution The 'Only x and y' Principle Bare Particulars Daffodils

33 37 40

45 51 58

62 62 65 67 69 73 75 80

84 84 85 91 95 99 107 109 114 120 125










5.1. 5.2. 5.3. 5.4. 5.5. 5.6. 5.7. 5.8.

129 132 134 139 146 148 158 166

Identity and Possible Worlds Worlds and Times Ontology The Properties of Stages Sortals and Recognition Real Possibilities Essence and Vagueness Temporal Parts and the Necessity of Origin



6.1. Structure and Objects 6.2. Components 6.3. Structure and Function 6.4. Artefact Identity 6.5. Petrified Forests 6.6. Identity and Survival 6.7. Colour, Camouflage, and Other Distractions

170 178 183 187 190 199 204



7.1. The Extensionalist Myth 7.2. Similarity 7.3. Critical Realism 7.4. City Blocks and Correlations 7.5. Similarity Circles

207 212 218 221 228



8.1. 8.2. 8.3. 8.4. 8.5. 8.6. 8.7. 8.8. 8.9. 8.10.

237 239 245 251 255 262 268 276 282 289

The Plot So Far Persons and Natural Kinds Mem,pry and Identity Criteria of Identity Dualism and Reductionism Fission Bodies Ownership Strict and Philosophical Identity Concepts of the Person



9.1. 9.2. 9.3. 9.4.

292 296 302 307

Connectedness and Continuity Amnesia Why Amnesia is Not a Problem Bodies Again

Contents 9.5. 9.6. 9.7.


Sleep Types and Tokens Death, Dualism, and Value

ix 311 315 3t9



10.1. 10.2. 10.3. 10.4. 10.5.

326 332 340

Branches and Persons Higher- and Lower-Level Concepts Nagel's Brain Forensic Conceptions of the Self Prospects and Retrospects

345 349





Eusa 33 Eusa sed, How menne Chaynjis ar thayr? The Littl Man sed, Yu mus no aul abowt that I seen yu rite thay Nos. down in the hart uv the wud. Eusa sed, That riting is long gon & aul thay Nos. hav gon owt uv my myn I doan remember nuthing uv them. Woan yu pleas tel me how menne Chaynjis thayr ar? The Littl Man sed, As menne as reqwyrd. Eusa sed, Reqwyrd by wut? The Linl Man sed, Reqwyrd by the idear uv yu. Eusa sed, Wut is the idear uv me? The Linl Man sed, That we doan no til yuv gon thru aul yur Chaynjis. RUSSELL HOBAN, Riddley Walker The dividing of reference consists in settling conditions of identity: how far you have the same apple and when you are gening onto another. W. V. QUINE, Word and Object, §24

Introduction is a voyage of a very special kind. In it the traveller is subject, for the most part, to constraints which would be intolerable to the social scientist, literary critic, or historian. For the appeal of much contemporary philosophy is to those of a highly rational temperament with a great respect for logical consequence. One rather irritating consequence of all this" is the appearance, on occasion, of pedantry, a tendency to literalness, and a close attendance to the details of language reminiscent of the fabled Vermonters. Smullyan reminds us of the story about Calvin Coolidge visiting a farm with some friends. When his group encountered a flock of sheep, someone remarked that the sheep had just been shorn. 'Looks like it from this side,' was the president's reply. In keeping a diary of my explorations of identity in the following pages, I have not lived up to the standards of Coolidge. Rather, I have tried to convey some of the excitement and interest of the topic by taking a relaxed approach to it. Although I deal with many of the topics that are of central concern to the theory of identity, I have kept technical matters to an absolute minimum. There is, however, more to say by way of introduction than this. Despite some respect for the three L's (attend to the Literal, to Language, and to Logic) any work in philosophy needs to be prepared to engage in theorizing. Everyone, of course, theorizes some of the time-even off-duty philosophers. But in this book I have presented a sustained example of philosophy as theoryconstruction, and have done so at the risk of occasionally not showing tremendous respect for the three L's. Theorizing in philosophy is not constrained by the techniques of science. For instance, there are few references to objective, repeatable experiments, except in the chapter dealing with some psychological issues. But there are many sorts of perfectly respectable kinds of theorizing which are responsive to the everyday and scientific knowledge which we possess, and some of these have a place in philosophy. To the extent, then, that this book succeeds as a piece of philosophy, it is a vindication of my view that theory construction, PHILOSOPHY



and certain kinds of speculative endeavours, are a genuine part of the business of philosophy. These remarks are not meant to exhaust the story of what philosophy is, or indicate any limitations on its scope. As our oldest academic subject, philosophy is also, perhaps, the most polymorphic. Further, contemporary philosophy as practised in most universities in the UK and North America is, for all its diversity, just one style of philosophy among others. It is not easy for dedicated practitioners of a certain style of philosophy to take this last remark seriously. However, I regard it as a limitation on my own work that my training and skills only pennit me to write one kind of account of the sikhts I have encountered. To the extent that theorizing is an attempt at giving explanations, the notion that philosophy involves theorizing-at least for some of the time-suggests that philosophy is concerned with explaining things. And given the connections between explanation and understanding, philosopby will be concerned with understanding. My initial question is: in what, if anything, does the identity or unity of a persisting thing consist? Related to this general puzzle are more specific ones, dealing with whether identity is, or is not, primitive and indefinable, and concerning whether the identity of different kinds of thing is itself different. Occasionally, although epistemology is not my main concern here, I also have to consider questions of how unity is detennined or discovered. In answering the initial question, I produce a theory that is meant to yield understanding of what the unity of a persist,ing thing involves, and an understanding of how we are able to recognize such unity. It is clear, then, that unlike some other contemporary thinkersNozick, for one-I do not restrict attention solely to questions of how something or other is possible. If the reader is convinced that some discoveries of interest are described in the following pages, then this will perhaps make plausible the claim that there can be systematic theorizing outside the sciences which does not descend into merely idle speculation. Someone who does dabble in philosophical theorizing is liable to fall foul of those austere souls who fear that too much speculation and too many thought-experiments lead us far from reality and, ultimately, into fairyland. The label analytical helps to suggest that philosophy is concerned with uncovering the nature of our existing beliefs and concepts rather than with developing new ones. If all the



analysts were after was a reminder of the need to hold fast to Russell's concern that we do not lose touch with the real world, then their worries would be perfectly justified. However, I hope to show that some flights of fancy simply help focus our thoughts on problems that are already real. The use of thought-experiments is no more than an enjoyable means for working on issues that even the soberest of analysts must recognize as important, and as demanding solution. At various points in the book I return to this issue, indicating in a conciliatory vein the relation between a particular flight of fancy and a real problem of analysis. In a less conciliatory vein, we might also observe that some enemies of thought-experiments simply reveal an astonishing degree of conservatism about what they will and will not discuss. Should we, for example, be constrained in our philosophizing by restricting our attention only to those theories which have the stamp of approval from contemporary science? Suppose that apparently sincere persons tell me that they know someone who is the reincarnation of a person long dead. Our best current science provides nothing which would enable us to make sense of such a claim in terms of physical or chemical models. But that science cannot vindicate the claim does not mean that we cannot make any sense at all of it, nor that we cannot investigate what difference it would make to our views if it turned out that reincarnation did occur. To refuse to think about phenomena lying beyond our present scientific ken is to adopt a stance that is not only blinkered but is ultimately alien to the very spirit of inquiry that drives science and philosophy forward. It is worth noticing that even the most sober of analytical philosophers are capable of occupying positions that are entirely alien to common sense. Quine is a notable example. For him, science is no more than an extension of common sense, and psychology is a branch of science. The theory of knowledge is, moreover, no more than a sub-department of psychology. So what does this sober analyst have to say about our knowledge of physical objects? These, Quine tells us in his early work, are not the basic items we might have thought. Rather, we have a theory that the world around us contains enduring physical things, and these things are themselves no more than the entities of theory-posits. As some kind of reassurance, Quine urges us to remember that to call a posit a posit is not to .patronize it.



Like all good philosophers, Quine shows commendable consistency in following his view to its logical conclusion. But in urging us to accept that enduring physical things are merely theoretical entities he has, in my view, left common sense, and any sense of reality, far behind. By contrast, the theories of identity put forward in this book are, I would maintain, far closer to the scientific, common-sense world-view (if there is such a thing) than the theories of Quine, or those of rival identity theorists. The reader who is initially hostile to my approach may like to bear this point in mind. Such a reader I would also urge to give careful consideration to my claims about, and examples of, conceptual development. If I am right in thinking that what is done under the label of conceptual analysis is often quite a different enterprise, then it is high time that this enterprise was recognized for what it is.

1 Identity and Continuity


Most students of philosophy are familiar with the dictum that you cannot step twice into the same river. In the two thousand odd years since Heraclitus (or whoever) made the observation, we have seen disappointingly little progress in solving problems about identity. My claims in the present work are fairly modest. One is to show how we can give a plausible account of identity that will cover someeven if not all--of the interesting cases. This will yield us a theory which gives us the identity conditions for many sorts of objects; for trees, mountains, and perhaps cats, even if not for bicycles and other artefacts. Some of the questions posed about identity, have no determinate answer. Of this kind are some questions about fairly humdrum objects and also some of the puzzles about personal identity. In some of these cases, at least, we can make practical decisions about what to do, hope for, expect, and think, even while leaving the question of identity unsettled. Showing just how we can do the latter constitutes a further aim of the work. It may seem that such a view is deflationary. For if we can come to practical decisions while leaving questions of identity unresolved, how can identity be what matters in such situations? We will see that, despite this deflationary possibility, identity does not become entirely discredited in my theory. Indeed, one large aim will be to show how identity is established in some cases, and in what identity then consists. Why identity matters, though, is that it is regularly associated with other things that matter. Such a view is bound to shock those who believe that identity itself is something of fundamental importance, a feature of the world that underpins, for instance, our naming and referential practices. I can only appeal, at this point, to such readers to keep an open mind for the time being. They will soon discover that the theory explained in this work has the virtue of resolving many otherwise intractable puzzles. Then they can, judge for themselves whether this benefit is worth the cost


Identity and Continuity

of giving up the view that identity itself is too basic ever to be capable of elucidation or definition in other terms. Identity, as defined in the majority of philosophical treatments of the notion, is related to another puzzling notion, that of similarity (or, sometimes, qualitative identity). We can reveal the connection between the two notions in informal terms thus: Similarity: If a is similar to b then something true of a is also true of b. Identity: If a is identical with b (that is, a=b), then everything true of a is also true of b. Instead of making the distinction in terms of truth and falsehood, we can make it-equally informally-in terms of properties (and relations) : Similarity: If a is similar to b then some property of a is also a property of b. Identity: If a is identical with b (that is, a=b), then every property of a is also a property of b.

Here, since we are talking loosely and pre-theoretically, a has a property if it has true of it some sentence ascribing a property or indeed one that assigns a relation between it and another thing. Understood in this way, similarity is a pretty broad notion. I do not normally regard my coal-scuttle as being at all similar to the still life that hangs in the sitting-room. However, both are similar, by the above account, through sharing the property of being in the sittingroom. Later, we will try to construct a more plausible notion of similarity, although this is no easy matter. Similarity is an important notion in the theory of knowledge. Our knowledge of the world is dependent on our ability to identify similarities among things. There are different views about what we should say about similarity. Perhaps, as Quine has suggested, we have some kind of innate 'similarity space', which enables us to classify the Heraclitean flux of experience into categories. Thus we are able to determine, for a newly encountered item, whether or not it is of the same sort as items we have met with previously. And, certainly, if we could not thus classify objects into sorts or kinds, it is hard to see how even the most elementary form of communication regarding our environment could get under way. Quine's theory of similarity spaces, however, is not the only, or the best, account of our similarity skills. As I will argue later, the world we live in may

Identity and Continuity


itself be the provider of similarities to us in the first place, and the existence of such similarities may be an important factor in making it intelligible to us. Whatever account we give of similarity, its importance is undeniable. Identity, likewise, seems to be a fundamental notion, one which-like similarity---combines this importance with diverse theories about how it is to be elucidated and what our standards of identity are. Some examples will reveal its claimed significance. Suppose, for a very simple case, I am telling a story about an incident I witnessed last week. One person who figures in the story is a woman in a stained trench coat whose name I do not know. However, to add vividness to the story I give her the name 'Amanda'. Now I can tell you various things about what Amanda did, where she was, and how she came to be, let us suppose, arrested. You can ask me if Amanda put up a struggle against the police, or whether her behaviour was such that she really ought to have been arrested. Here, each time either you or I have used the name 'Amanda' and its correlated pronoun, we have taken these terms to refer to one and the same person. And to speak in this way about 'one and the same person'- is to use the notion of identity: identity is, we might say, the property of being one and the same. Not just proper names and pronouns, but arbitrary symbols and certain descriptive phrases get used quite regularly in this way: so that each time the same expression is used, one and the same thing is referred to. So it seems that our very apparatus of naming things, and referring to them subsequently, presupposes the existence of identity, and involves the concept of identity. For the time being, let us allow the strong claim that the reference of names and descriptions does in fact require this presumption of identity, and continue to look at the sort of argument that tries to establish the centrality of the notion. Think again of individual things, objects. As already pointed out, the objects we talk about in daily life belong to well-established sorts or kinds-trees, goats, fields, bicycles, desks, cranes, fence posts, swifts, and so on. Each of the nouns used in the list given seems to apply to items which we can count. So, as well as calling such nouns sortalsJ or terms of divided reference, we could also call them count-nouns. I Terminology here is I Griffin defines count nouns thus: 'A general term, "T" is + count if "There are n Ts ... " makes sense, where "n" is a variable taking numerals as values; otherwise "T" is-count.' (Griffin [2], p. 23.) Although such a characterization would no


Identity and Continuity

various and-without wanting to impose one nomenclature on the reader-I will mainly use the term 'sortal' to apply both to the nouns and to the related concepts. The important thing about sortal general terms, and the associated sortal concepts, is that their exercise seems to be associated with principles for counting and individuating the items to which they apply. So the items associated with a sortal (term. or concept) thus constitute a collection of things that may be counted, things like goats, trees, and cows, where it is fairly clear, therefore, where one such thing ends and another begins. 2 Whatever we call such terms, someone who could not tell a swift from an eagle, or from a fir-tree, would show a poor grasp of the use of the term 'swift'. More significantly, someone who could not distinguish one swift from a flock of them, or who thinks that a flock of swifts is just another swift, would be someone who has not grasped the use of the term 'swift' (nor the associated concept). And such a person would never learn to count swifts, if unable to grasp the conditions of identity for swifts-the conditions under which one swift ends and another begins. Of course, grasp of the sortal 'swift', and the ability to count may not go hand in hand. We can imagine someone whose best attempt at differentiating swifts went 'One, two, several, many, lots and lots .. .', and this would not be a person skilled in counting. But such an example does not undermine the claim that sortal terms do have principles of counting associated with them. Sortals in fact have all kinds of interest for the identity theorist, and we will be looking at some aspects of this in due course.

doubt be roughly adequate for defining sortals, Griffin prefers to distinguish the latter from count-nouns, giving this modification of a criterion originally due to Sheehan: 'A term "A" is sortal if there can be cases in which "A" provides, without further conceptual decision and without borrowing other principles of individuation, principles adequate for counting A's.' (Griffin [2], p. 43.) Thus 'thing' is a count-noun, by these definitions, but not a sortal. As is clear in the text, I am happy to accept that sortals, rather than count-nouns, are of interest to theories of identity. Sortal and count terms show an interesting difference from mass terms, the latter being, in Hirsch's terminology, dispersive: 'A term like 'wood' is dispersive because any stretch (quantity, bit) of wood will extensively overlap numerous other stretches of wood that make it up.' (Hirsch [2], p. 42.) Hirsch is thus able to define sortals as nondispersive terms. 2 As stated in the epigraph from Quine [2], § 24. Although it may not seem necessary to distinguish sorts from kinds, I treat sorts as forming the more general category. Thus, one important variety of sortal terms is the set of natural kind terms.

Identity and Continuity


Not only is our understanding of identity, of someness and difference, associated with our understanding of the world as containing individuals, themselves of various sorts, it is also associated with other concerns. One of the most significant of these, perhaps, is our concern with persons and our care about their future, their pasts, and their responsibilities. Some people, late in life, still feel troubled by guilt about actions they carried out when children or teenagers. Others, often religiously inspired, claim an interest in a future that continues after their bodily death. Our punishment of offenders, our criticism of our own, or others', moral failings are based on an understanding of personal identity which goes unchallenged for the most part. Yet as soon as we ask the right questions, we can find ourselves puzzled. Am I the same person I was as a child? How could I be the same person I am now after the destruction of all my body? If I had a brain transplant after comrnittin~ a crime, would'!' (if it is still me that we are talking about) deserve punishment for that crime? Not only has this topic attracted a literature all to itself, but it poses problems that seem almost as great once we have studied other cases of identity as they seemed when we first started. One virtue of the theory of identity ventured here is that it makes some of the questions about personal identity seem easier to answer than we might have thought. Unfortunately, for some of the problems, as we shall see, there remain no determinate answers nor is there-I would argue-hope of achieving determinancy. Let us conclude this introductory section by looking at the formal principles of identity. Two of these suffice for all the logic of identity. The first, the principle of relexivity of identity, simply states that'everything is identical with itself, in symbols. Reflexivity: 'v'x[x=x]

The second simply repeats the informal statement already given: that if x is identical withy, then everything that is true of x is also true ofy. This principle is given many names in the literature, two of the most common being Leibniz's Law, or the principle of indiscernibility of identicals. In symbols, we have Indiscernibility: 'v'x'v'y'v'F[(x=y):::J (Fx:::J Fy)]

I have given this in its second-order formulation partly because it seems natural to do so, and partly because it avoids certain formal


Identity and Continuity

niggles. 3 In this form, we can use the two principles to derive a third, important principle. Unlike the uncontroversial fact that identity is symmetric (if x = y theny = x) and transitive (if x = y then y=x) and transitive (if x=y and y=z then x=z), this further principle states the identity of indiscernibles, and is controversial enough to have been dismissed by Peirce as 'all nonsense'. The principle tells us that if everything true of x is also true ofy, then x is one and the same as y, in symbols: Identity of Indiscernibles: VxVy[VF(Fx::::l Fy) ::::l (x =y)] It is tempting to suggest, in view ofthe difficulty some people have in accepting the third principle, that classical second-order logic enriched with identity is in much the same position as first-order logic given the Boolean definitions of the conditional and conjunction. The system, in each case, works well. But enough questions are raised to point to a discrepancy between the logical systems we use on the one hand and the system of reasoning encoded in our natural language on the other. Fascinating though such a topic is for investigation, I ignore it in this work.


It is one thing to know the logic of identity, another to look for those conditions of identity that apply to the case of correctly reidentifying objects. Theorists who accept the principles of reflexivity and indiscemibility have fallen by and large into two distinct camps 3 For example, we do not need to consider the problems posed by adding indiscemibles (which may be distinct) to a first-order theory. For further comments on the fonnal principles, see the standard treaonents by Wiggins and Griffin. Ishiguro [1] gives more infonnation on Leibniz's Law. Sometimes indiscemibility is expressed in either of the following fonns, where the schemes are valid for all instances: (0 (x=y & Fx)::>Fy

(H) x=y::>(Fx=Fy)

(0 is obviously equivalent by truth functional logic to the version in the text. (H) is easily derived from that version given the principle ofrefiexivity. For suppose that x = y, but ~ (Fy::> Fx). Substitution for F yields ~ (x = y::> X = x), that is, x = y while ~ (x = x), which latter is a contradiction. Hence x = y::> (Fy::> Fx), which establishes (H).

Interestingly, as Quine shows in Quine [3], we can derive all the properties of identity from one principle, Wang's Law: Fy=3x(x=y & Fx).

Identity and Continuity


over the matter of how to reidentify objects-we can call them continuity theorists and indiscernibility theoristsJ respectively. The problems come about due to a phenomenon known to, and celebrated by, Heraclitus himself: change. If doors that were painted red never changed their colour, if bodies didn't change their cells, if rivers never changed their water or their course, if cameras, bicycles, and houses never needed new parts, then we would h~ve a duller world and far fewer problems about identity. Luckily, though, the world is not dull. Change is with us all the time, and this is the source of a great many problems. Perhaps, whatever the logical laws of identity state, an important feature for the identity of one persistent thing is its continuity in time and space. The continuity theorist can point here to the celebrated case mentioned by Hobbes, supposedly about the ship of Theseus. 4 After sailing from Athens, this ship finds itself gradually subject to replacement: plank by plank, nail by nail, sheet by sheet, the entire hull, masts, rigging, and so on are replaced by new versions which themselves look fairly similar to their originals. In the end, not a single dowel, not a single strand of hemp survives from the original ship. Yet all the time, the ship has been plying its trade, going to sea, colliding with jetties-in shon, functioning normally as a ship throughout the period of replacement. What is the right thing to say here! Has Theseus' ship been repaired or replaced? If the latter, then when did the original ship go out of existence-when the very first repair occurred, or only after a cenain amount ofthe original material had been replaced? If it has not been replaced, but has stayed the same all along, then it seems that one and the same thing can be two quite different collections of masts, spars, planks, and sails. Pondering on this question, we see that, for a stan, the logical principles already given do not settle the matter of identity for this kind of case. This may seem a disappoinunent, until we recognize that logical principles seldom supply legislation for conceptual puzzles: their ability to codify agreed inferences makes them hospitable to the expression of divergent theories. Thus those who think that the replacement of pans leads to the existence of two distinct ships can agree with those who think that one ship survives all the changes: indeed, failure to 4 The puzzle originated in Hobbes [1], pt. II, ch. 11, although, as Wiggins observes, Hobbes no doubt found the idea in Plutarch (see Wiggins [9], p. 92).


Identity and Continuity

agree on principles like Leibniz's law and indiscemibility would make the disagreement about sameness and difference assume a new character. It would become, at least in part, a disagreement about the meaning of the term 'identity'. Faced with the problem of change, a theorist wedded to the notion of continuity may point out that typically the concrete objects surrounding us are persisting things-continuants, to use Wiggins's word. The term 'concrete' is used in the last sentence in the philosopher's, not the builder's, sense. In the next chapter, more will be said about what this sense is. Persisting things can change; some changes are greater than others. But the changes to Theseus' ship are not such as to turn it from a ship into something that is not a ship. So perhaps the best thing to say here is that we have witnessed spatio-temporal continuity of one enduring ship. On this approach, we can still distinguish the ship at one time from the ship as it is at other times. Such a stage, or temporal part, of the ship will differ from other stages in terms of the collection of planks constituting it. Thus, one same thing, the persisting ship, has a number of different stages we can allocate to it. Let us suppose that we can make these stages long or short in duration, but that relative to the total history of the object such stages are pretty short-lived. The replacement of any part of the ship by a new part could be the point we choose for transition from one stage to the next. We could then think of the ship as a series of stages each one of which is itself a distinct collection of ship-parts. We can make this clear on a standard spacetime diagram depicting a portion of the ship's history (Fig. 1). Notice, of course, that this is only one of indefinitely many ways of partitioning the ship into stages. And there will still be changes during the stages: chips and dents and scratches, weathering, flaking of paint and tar, not to mention a host of minor changes all the way down to molecular and atomic levels. However we select our stages, we have here an attractive way of dealing with sameness and difference. The ship itself is more than one collection of parts. Moreover the collection of parts that constitutes the ship changes from time to time. The differences among stages plot the visible changes in the history of one enduring and continuous thing-the ship. Looking at the case of the ship in this way starts to make some sense of our second principle---that of indiscemibility. It is because we can trace continuants through time, and assign diverse stages to

Identity and Continuity



Fig. 1.

one and the same persisting object, that we can both allow for the phenomenon of change and at the same time insist that if a is the same as b then any property of the one is likewise a property ofthe other. In the case of our one ship, being constituted of collection Cl at one time and of C3 at a later time are properties true of it and nothing other than it. Sameness and change thus become incorporated into a comprehensible account of the ship's history. 1.3. CONTINUITY AND INDISCERNIBILITY

The continuity account is not the only one we can give here. As already pointed out, there are two different camps into which people divide on this issue; if one camp contains the continuity theorists, the other contains theorists who take indiscernibility as a sufficient condition of identity. Of course, these indiscernibility theorists are justified by our third principle; they even take it as the fundamental law of identity, rather than as a consequence of the others. Our newly introduced theorists deny that continuity has any central role in settling identity questions. Even though they admit


Identity and Continuity

that our world consists of relatively enduring and changing objects, they find continuity alone neither necessary nor sufficient for identity. Let us give some content to this disagreement by considering another space-time diagram (Fig. 2).5 This time we are tracing the history of an entity called-quite arbitrarily-'Bill'. Mter coming into existence, at all times up to and including t I Bill is fairly stationary in the region of PI. Immediately after El' however, Bill makes a sort of instantaneous 'quantum jump' to P2 where he remains for the rest of his life. Bill's history appears to be discontinuous in space (though not in time); and--of course--we are not used to such happenings in regard to objects in the macroscopic world. None the less, we seem to be able to make sense of the idea that we should assign these lengthy discontinuous stages to the history of just one thing. To complicate things, we can consider the history of some other entity, perhaps the Arnanda of our earlier example. Oddly enough, Arnanda occupies position P2 for some time right up to tl' whereupon she makes a quantum jump to place PI. We can thus


'I Bill

L------------------P PI


Fig. 2. 5 My version of the indiscemibility theorist's case is based on the discussion in Brody [3].

Identity and Continuity


complete our earlier diagram with the lifeline of Amanda, recognizing that her discontinuous existence in no way prevents us from assigning both life segments to the one entity (Fig. 3). Now consider what the continuity theorist has to say about this case. There are two continuous space-time paths shown on the diagram. The one at PI has a Bill phase followed by an Amanda phase, while the other has an Amanda stage followed by a Bill stage. If continuity in space and time is what really counts for identity, then, however different the properties of Bill and Amanda, we have to recognize that one same thing can begin its career with a Bill phase and then continue for the rest of its history with an Amanda one.



IAmanda 1


I 11


1______ ------1


I Amanda



I ~------------------------p

Fig. 3.

Put in these sketchy tenns, the continuity view of the matter seems distinctly implausible. So perhaps it would be wise to look at another example to see if we can sharpen our intuitions. However, the best cases for this purpose are--to put it mildly-somewhat fictionalized. One that we can try to explore concerns an unfortunate dog. Rover,. who is supposedly the victim of a NASA


Identity and Continuity

experiment. The plot is as follows. 6 Having been sent on a pioneer mission to Mars. our terrier returns to earth where, to everyone's surprise, a nasty and untoward effect of Martian radiation begins to show itself. In the isolation unit in which he is kept under continuous observation, Rover begins to manifest a number of slow changes. Over the weeks he gradually transfonns, rather like a caterpillar in its pupa stage--only to a far greater extent. Within months our dog has been transfonned into a mass of cells, totally without structure. Within the cells, even the constitution of chromosomes has changed so that the lump of matter that remains does not even have any of Rover's original DNA. Using a new name to designate the amorphous mass into which Rover has been transfonned, let us call the blob 'Clover'. Unlike the preceding case, we are not here faced with instantaneous change. And Marjorie Price, in putting forward this grisly story, is taking a line in favour of continuity. Her claim is that the example poses a difficulty for those who maintain that what is required for identity is continuity under some same sortal; for Clover has none of the interesting sortals true of it that were originally true of Rover. The mention of 'interesting sortals' is meant to preclude the objection that Rover and Clover are both things, objects, occupiers of space; none of the latter are what Wiggins would call 'substance sortals', and certainly those who make much of sortalcovered continuity do not have such uninfonnative sortals in mind. 7 So Price's view is that Rover and Clover are indeed identical, although they are not the same dog, the same animal, or the same anything else of an interesting sort. fi So writes Marjorie Price in Price [I]. For a thorough discussion of the case, Brody [3] is worth consulting. 7 For Wiggins, sortals provide the most privileged and fundamental answers to questions of the form: what is x? As well as obvious cases where sortals apply to an individual throughout its entire history-such sortals being, in Wiggins's treatment, substance concepts-there are two special cases worth noting. First, there are restrictions of some underlying sortal, where the restriction is concerned with a phase in the history of an item: thus we have 'child' as a restrictioIi of the substance concept, 'human'. Second, there are concepts and terms that seem to lack individuative force, and are thus too unspecific to count as interesting sortals, or even as sortals at all. Examples are 'entity', 'thing', 'space-occupier', and so on. 'Uninteresting' sortals, however, seem just to be extreme cases of fairly obvious disjunctive sortals (e.g. 'spouse' as meaning 'husband or wife'; 'insect' as meaning 'wasp, or bee, or dragonfly, or .. .'). The most sensitive discussion of these tricky matters is to be found in Wiggins's work, especially Wiggins [9].

Identity and Continuity



The example, however, serves well enough to show how those not committed to continuity as a sufficient condition of identity can argue that Rover and Clover are distinct things despite their continuity. Rover and Clover do, of course, sit on segments of one continuous world line: so does a living butterfly and the remains of its skeleton; so did Bill and Amanda. The indiscernibility theorist can argue in some, or all, of these cases that distinct things sit on these segments, notwithstanding the continuities. And the basis for the claim of distinctness is the lack of common properties: Rover and Clover are discernible; they are thus two rather than one. And the indiscernibility theorist can maintain this while admitting that there is no clear way of knowing just when Rover ceased to exist and when Clover started to exist. It has not been my intention thus far to defend one or another of these positions. As the theory of identity unfolds in later chapters, we will see that there are interesting things to be said about continuity and about discontinuity. At the moment, I am concerned to impart the flavour of some of the central puzzles in the area, and to establish the sorts of positions that can be taken towards these puzzles. Later, we will see that some of the puzzles are more apparent than real, and that what appear at first to be quite baffling problems will yield to essentially simple treatment.


In 1971 Derek Parfit published a defence of the claim that survival may often be more important than identity. 8 What does he mean by 'survival'? Is it not just identity under another label? In Parfit's early treatment, we can think of survival as a kind of default option: it is the feature we default to when conditions for identity are satisfied but something else rules out an identity claim, or when the matter of identity itself is indeterminate. It is clear that we can ask of 8 See Partit [1] and [2]. Although I borrow the term 'survival' from Partit's early work, he subsequently dropped it. In Reasons and Persons he simply writes of psychological continuity and connectedness as being the relation that matters for psychological unity. This is convenient, for my own account of survival may well not be one that Partit would endorse. In subsequently reconstructing his view on persons, I will continue to use the term 'survival', understanding it in the way defined in the present book. I hope this terminological device does not cause unnecessary confusion.


Identity and Continuity


Theseus' original ship whether or not it survives the replacement of its parts; and we can likewise ask if Rover survives as Clover. Parfit's point is that we can make sense of survival even where identity is out of the question (for example, where the original ship survives twice over), or where we are unable to determine the right thing to say about identity. We can thus have survival without identity, even though we may not have identity without survival. That we cannot have identity without survival is rather easily established provided we accept the logical principles already introduced and make the assumption that everything survives as itself. Suppose, now, that a = b; then, it must be true, by our assumption, that b survives as b; so, by the indiscemibility of identicals, the very same thing-surviving as b-must be true of a. So if a is one and the same as b, a survives as b. It is therefore more interesting to examine the converse case. Let us start with a case of the second kind mentioned above--one where we are unsure what to say about identity, but still seem able to make a claim about survival. Parfit considers the intersting case of resurrection. 9 Although personal identity is a topic for special treatment later, we can dwell for a moment on this case without raising all the special considerations demanded by study of persons. Suppose, then, that we think of the Great Day of Judgement in the following terms: God creates replicas of all those who have ever lived, in order to pass judgement on them. These replicas, let us suppose, are accurate down to the finest level of biological and biochemical description. Given that an enormous number of years may intervene between an individual's death and subsequent replication, we can see that there is a real problem over whether to count the replicas as identical with the originals. If I am numerically one and the same as the replica God makes of me, then, of course, it will be I, and no other, who stands judged on that fateful occasion. On the other side, if I am not identical with my replica, then it is not I who stand to be judged, but someone extremely similar to me, who is judged in my place. Suppose, at least for the argument, that we cannot decide between thse two situations: we just don't know how to cast our vote on the identity question. Parfit's interesting suggestion is that we should just forget this problem, and concentrate on the survival 9

The discussion in the text is based on Parfit [2].

Identity and Continuity


question instead. An item can not only survive as itself but can survive as or in something else. Given our understanding of genetics, for instance, it is no wild metaphor to suggest that parents may survive in their children: parental genes do indeed get passed on to offspring. With this in mind, we can start to see how Parfit is able to prise the identity question from the survival question. If identity is the important thing for survival, then our puzzlement over the right answer in the resurrection case leaves us puzzled over what to say about survival in that case. But shift the focus, and see survival as the crucial thing, and we can start to think of resurrection by replication as being as good as survival. There is no doubt that many people find Parfit's account of survival obscure. One reason for this is that we think that answering identity questions is a way of getting clear about survival. His suggestion, correct in my view, seems to be that survival may be the more primitive notion. Indeed, as will become clear, I think that on one approach we can use survival to explicate identity, just as Tarski found that the apparently primitive notion of truth could be explicated by a still more primitive relation of satisfaction. 10 Moreover, the analogy with Tarski has a further feature. Survival, in what we may call its genuinely relational sense, underpins and explicates identity, where identity is not genuinely relational. By 'genuinely relational', I mean some relational feature that links distinct things, whereas identity as a congruence relation simply links one thing with itself. Likewise truth seems to be a nonrelational property of sentences or propositions while satisfaction, as understood by Tarski, is a relation between quite distinct things---open sentences, say, and sequences of objects. For those who do think identity is the primitive, or fundamental idea, with survival depending on it, a great deal of charity is needed to make sense ofParfit's radical alternative. The case of resurrection is not by itself going to swing their intuitions into some new alignment. In the third chapter, I will give a quite precise account of what it is for one thing to survive in or as another, but for now I will 10 This can be found in Tarski [I]. There is a certain air of hocus-pocus about Tarski's definition of truth, since it depends on a technical uick involving a truthlike notion: satisfaction, after all, is simply the relation of being. true of. My account of identity involves no fonnal tricks of the son Tarski plays-and so, as it seems to me, gives a satisfying account of identity in some cases. For a helpful introduction to the Tarski machinery, see Wallace [I].


Identity and Continuity

make do with some further intuitive examples which may-I hope-establish at least the plausibility, if not the correctness, of Parfit's line. Think, for example, of the hope that authors entertain of immortality through their works. Here there can be no question of identity between author and work, yet we can make sense of survival to some extent. Notice that, unlike identity, survival comes in degrees. We could never talk of identity to some extent: items are either self-identical or not (only items like the empty set are not selfidentical); if' a' and 'b' are different names, then' a = b' is either true or not (but not true to a degree). Switching from people to other sorts of things, we can talk about the relative survival of buildings, paintings, pieces of music, and so on. How much of Vivaldi, we can ask, survives a transcription by Bach, or a subsequent transcription to synthesizer? After an ambitious series of conversions and extensions, very little of an original house may survive, but that little which does survive may be of importance to the occupiers. So even survival to a small degree can be of significance in certain circumstances. Those who perhaps have qualms about the resurrection case may, by considering cases like these, start to feel that talk of survival does make sense at least some of the time where talk of identity would be inappropriate. We can now elaborate the case of Theseus' ship to provide an example of a case where we know identity is out of the question. Here we are looking at a case of fission, and the point concerns the existence of rivals for identity with some original. In honour of the champion offlux, let us christen the ship that originally set sail from Athens the Heraclitus. That good ship stayed afloat throughout its various refurbishings, and by the end of its career, we find it more or less structurally the same as the sparkling original, although a deal shabbier and mouldier. Now imagine that while the repair work was under way, some eccentric of the ancient world set about collecting together every item discarded during the repair process. Over the years our plank-hoarder, no doubt at great difficulty and at some expense, built up an increasing collection of original parts from the old Heraclitus. At least, the project of constructing the Heraclitus as she originally was can be realized: a ship grows on the stocks, not only structurally identical to the H eraclitus, but consisting of all the original materials. Eventually, this extremely shabby replica puts to sea, competing for the honour of identity with the original. Just to keep our minds minimally clear, let us name the plank-hoarder's

Identity and Continuity


ship the Heraclides (after an obscure head of the Academy). We can imagine a time when both ships lie at anchor together: for those who sympathise with Parfit, we can use the vocabulary of survival: the old Heraclitus seems to have survived twice over. The identity theorist, of course, is interested in the question of which ship we are to identify with the original: the Heraclides, for all its tattiness, can claim a pretty pure pedigree; while the Heraclitus has shown continuity under the name and throughout the various refurbishments. It is clear that although the Heraclitus may have survived twice over, she is not identical with both surviving ships. For two things cannot be one thing. It is perhaps best to consider this problem by comparing the situation just described with two others. We can think of the cases in terms of a diagram (Fig. 4). The dotted lines in situations (a) and (b) represent collections of ship stages during the repair process, while the alternate dots and dashes represent the plank-hoarder's ship. Our natural response might be this. We think in case (a) that the Heraclitus is the same throughout the repair process: the figure represents the life of one ship. In (c) the Heraclitus seems to be the same ship as the Heraclides. In this case, the ship in question has undergone what we might call afunctionally discontinuous existence. This is a common kind of life for some artefacts-tents, collapsible boats, garden sheds, and so on are often constructed so that they can be disassembled. Big ships seldom are these days, although many of the steamers of inland Africa were made in this way. So we can probably make sense ofthe Heraclitus



'. ".


/ ' / Heraclides ... ...........



H eraclitus •




Fig. 4.


Identity and Continuity

ceasing to function as a ship for some time, and then being rebuilt to undertake its functional life once more. The case in the middle, case (b), is the one that gives us the greatest headache. For here we do not seem to know what to say. Of course, we can try to come to an identity decision in such a case. And such a decision can be important (for settling legal matters about insurance or ownership, for example). Parfit's position here is to deploy the notion of survival independently of our identity decision. Suppose that we give the judgement for identity in favour of the continuously functioning ship (from solid to broken line). Then Parfit can still point out that the original Heraclitus survives as the Heraclides although not identical with it. And this seems entirely right. Our eccentric has assured a double success, a success which is not captured by our verdict on identity.


We have a stubborn tendency to think that all sorts of problems will yield to suitably dedicated thinking. Thus it comes as something of a surprise to find that in some of the examples we have so far considered, the issue of identity may stay forever unresolved. This is not the case in all of them. For example, it is easy enough to think of the Theseus ship case as one in which we can probably be persuaded to give our verdict in favour of the continuously functioning ship. In the resurrection case, by contrast, it is not easy to decide for or against identity. Of course, if what is important for practical decisions is survival rather than identity, we lose one fairly significant incentive for worrying about the identity question. In the third chapter, we will look again at Theseus' ship with the issue of importance in mind. In the fourth chapter, we will defend the informal account sketched here in which the presence or absence of rivals seems to make a real difference to our verdict on identity. At the moment, it suffices to point out that importance is relative. When asking whether or not identity is important, we want to know the answer to the question: important for what? As we will see, survival can be important for identity itself! Let us suppose, however, that we are satisfied enough by the survival story, so that we lose interest in the matter of identity in some cases. We may thus come to regard the identity issue as

Identity and Continuity


indeterminate. It is not necessary to take a strong line on the matter of indeterminacy here. One legacy of Quine's work is that many contemporary philosophers regard genuine indeterminacy as a real possibility. But we need not commit ourselves at this stage to any kind of Quinean indeterminacy about identity (which would involve admitting the existence of ontological relativity).1I Rather, we can just keep things at an epistemological level : suppose we do not know what the answer to the identity question is when faced with situation (b) in Fig. 4. Then whether our ignorance be due to some inscrutability in the very notion of identity itself is not something about which we need currently venture an opinion. Nor should we make the mistake of regarding our knowledge about survival as always more determinate than our knowledge about identity. In the case ofTheseus' ship, we certainly seem to be able to speak about the Fig. 4(b) case as one in which the original ship survives twice over. But we may not always be so lucky. There may well be occasions where our ability to pronounce on the survival question is no greater than our ability to pronounce on questions of identity.


The conception of ordinary things as continuants makes sense of one prevalent feature of our experience. This is that how a thing is at one time very much depends on what happens to it at others. Objects carry their history on them in various ways. High-level items, like persons, not only carry the scars of cuts and broken bones, but also have memories as depositories of their history. But the scratches on my desk, or the old ink-stains that have never quite faded, are themselves a testimony to the desk's past existence: it is an object with a history. Let us think of things now in terms of their stages. Each extended object (extended through space and time) consists, so we suppose, of successive stages. This supposition, of course, is open to question: Hume regarded our belief in persisting objects as a vulgar confusion on our part between the ideas of relation and of identity. 11

Quine's accounts of these matters are to be found scattered through Quine [2]

IUld [7].


Identity and Continuity

But let us not give in to such doubts at the moment. So the successive stages of my desk (however we choose to divide it into these) will show a regular pattern of relation to each other. Suppose, for example, we think in terms of hour-long stages. It would be wholly baffling if from one hour to the next the desk gained or lost legs in a random way; if scratches present at ten in the morning had disappeared by lunch-time, but reappeared by tea-time; if the large drawer on the left had changed places with the small drawer on the right an hour after we had checked their position, and so on. Of course, we could tell a story to make sense of all these happenings. For example, maybe someone takes great care to make scratches on the desk every morning, notes their position, polishes them out, and then makes them again in the afternoon. Or maybe a carpenter modifies the drawers and their housings at regular intervals. These explanations would immediately demystify the situation. Why is this? The reason is that, without such accounts, we would be in a genuine difficulty when trying to explain the sort of changes imagined. Scratches do not appear, disappear, and reappear without some responsible causal agency. The story of the compulsive scratcher and polisher supplies just such a suitable causal account. Nor do desks gain or lose legs, or have their structure modified in other ways without our being able to assign appropriate causal explanations. If my desk is clean and unmarked in the morning, I expect it to remain so--unless something happens to it. If I return to my office at lunch-time and find scratches and inkstains on it, I do not just suppose that somehow the desk-stage I witnessed earlier in the day has, fortuitously, given way to a deskstage of a scratched and stained kind. Rather, I presume that something has happened in the interim. What all this amounts to can be expressed as follows: (1) The stages of a persisting object succeed each other in a

lawful and regulated manner. (2) If a later stage manifests certain differences from an earlier stage, then something has affected the earlier stage, or some intervening stage, and that thing, whatever it is, is responsible for these differences. The differences and changes here are, of course, not merely Cambridge differences. In the 'Cambridge' sense, my desk has

Identity and Continuity


undergone a change if, say, a certain car is now further away from it than it was a few moments ago. That is, a Cambridge change occurs simply if some proposition about my desk changes in truth value. Such changes affect objects all the time: but they are not the sorts of changes we can discern just from inspecting the object concerned in isolation from everything else. It is a philosophically difficult matter to give an account of properties and changes which captures just what is involved in a 'real' rather than a merely 'Cambridge' change. But this problem is not one for tackling here. All my talk of properties and changes, unless explicitly qualified, should be taken to be about 'real' properties and changes. Our observations, or rules of change, (1) and (2) above, go some way towards meeting Hume's scepticism about identity and relation. For the hypothesis that objects trace extended continuous paths through space and time seems to be the simplest one that accounts for the general coherence underlying (1) and (2). None the less, as we shall see in Chapter 4, it is possible to imagine a more Humean world in which (1) and (2) are satisfied while objects fail to be continuous. Leaving this to one side, however, let us consider the related question of survival. Is there, in one persisting object, survival of any of its stages in, or as, others? As we have already seen, survival of one thing as itself may be trivial, while survival of one thing as something other than itself is non-trivial and possible. Think again of the case of the parent and the child. When we say that the parent survives in the child, what do we mean? And would our account here be able to throw any light on the question of the survival of one stage of a persisting object as a later stage? We can take these questions in turn. When we say that a parent survives in a child, we suggest, by our use of the term 'in', that the survival is not to as high a degree as is possible. Likewise, suppose I wake up one day with the memories of someone else--Jane Austen, for example. Then we would tend to say, if we made a survival claim at all, that Jane Austen survived in me, but not that she survived as me. Remember that survival comes in degrees, and we can have survival of one thing in another thing to a pretty low degree. So when would we be likely to say that one item survived as another item? We can construct clear cases of this. For example, a child designs a marvellous bridge using a plastic construction set. Her brother, inspired by her work, models the bridge exactly using wooden blocks and glue. The original bridge is


Identity and Continuity

thereafter taken apart and the plastic components of which it was made are used for the next creation. Has the child's original bridge survived? The answer to this question is that it has survived as the wooden bridge. Those who are happy with the terminology and already feel that they can make sense of survival as in these terms will need no reassurance from further examples. But the fact that it seems natural to me to speak of survival in this way may be simply eccentric. So this example may be taken, by those who cavil over my use of language, as stipulative. Likewise, the same readers can take my remarks about surviving in as equally stipulative if they find my linguistic intuitions out of step with theirs. The example of the toy gives us all we need for sketching an account of conditions for survival. Note that the following two conditions are satisfied in the situation described: Structure condition: the brother's copy is structurally the same as the original; that is, it has components of the same size and shape as those of the original arranged in the same relative positions; Causal condition: the causal account of how the brother's model was constructed makes reference to his sister's original model; in particular, his version was obtained from hers by being copied, albeit in a different material. There are clearly features of these conditions that require more investigation. For example, in the structural condition what do we mean by 'component'? And what difference does relative size of the components make to survival? Would topological deformations also preserve structure? Of the causal condition, we can ask about the varieties of appropriate causal story, or about how accurate the copying process need be, and so on. We will tackle these detailed questions in due course. In the meantime, we might wonder if there is anything more to survival than these two conditions. We might ask if there are prospects for the survival of the original model that would result in an even higher degree of survival for it than in the imagined case (stopping short of preserving the original itself). Clearly a higher degree of survival would be obtained if the brother used the same sort of plastic bricks for his model that his sister used in the original. Of course, they need not be exactly the same bricks: indeed, we would face ourselves with a decision on

Identity and Continuity


identity if we imagined that the original model was taken apart, then the same bricks used in the same positions for constructing a bridge. One natural way of describing this latter case would be to say that the bridge had been dismantled and then put together again. What we are looking for is a case where there is no identity claim to be made, but none the less a really high degree of survival. Use of similar bricks in the same arrangement gives us this case. Thus we can suggest a third condition on survival that takes account of material similarity: Matter condition: one item survives as another when, in addition to the conditions on cause and structure being satisfied, the item as which the original survives is made of material similar to the original.

In both this and the structure condition, it does not really seem to matter whether we speak of the same matter or structure or of similar matter and structure. This is because sameness of such things'itself can come in degrees (for example, we can speak of 'much the same matter' and 'almost the same structure'). The conditions obviously prompt questions. What happens if we vary them to certain degrees: do we then give an account of survival to various degrees? Does the reference to sameness of structure and sameness of matter mean that survival depends ultimately on identity-albeit identity not of objects but of more abstract things? Even at the moment, we can suggest that, of our three conditions, only the two concerning structure and cause are necessary for survival. For, since one thing can survive as another, we can hardly insist even on similarity of matter in all such cases. Further examples in Chapter 3 will show that this proposal is along the right lines. By contrast, all three of the conditions taken together seem to constitute a set of sufficient conditions for survival. If one item possesses lots of material similar to that possessed by another, if the second item has that material structured in the same way as it is in the first, and if the first item participated causally in bringing about this sharing of structure and matter, then the first item survives as the second. As the theory of survival unfolds in the following chapters, we will tackle the questions raised here in the hope of answering enough of them to show that the theory is interesting. Looking now to the case of the parent surviving in the child, we can start to get some idea of how our conditions fare under various


Identity and Continuity

kinds of weakening. If the model bridge case is a paradigmatic example of survival under optimum conditions, the parent-child case is probably one of the weakest cases of survival recognized in literature and common sense. Parents who think they will not wholly die thanks to the preservation oftheir characteristics in their offspring believe something we can all make some sense of. In tenns of our conditions, we can try to make clear just what this hope for survival is based on. The structure condition suggests that parental structure is preserved in the offspring. Can we make sense of this? We need to think of two sorts of structure here: first, structure as a physical characteristic, and, second, structure in a more extended sense. As to physical characteristics, it is clear that offspring bear various resemblances to their parents: these structural characteristics are coded for by genes and therefore to be expected; and even those bits of structure nonnally hidden from casual inspectionlocation and size of the liver, for example, or pattern of bloodvessels supplying the retina-will be genetically determined. More difficult to deal with are those characteristics of personality, mood, and behaviour which parents recognize both in themselves and in their children. Here we need to think about the structure of personality and of mind-something that requires an extended notion of structure to be deployed. Later, I will suggest that it is particularly helpful to think of the mind as having a structure, just as a physical object has a structure. Just as the structural invariances in someone's facial features enable us to recognize him or her after a change in hair-style, after gain or loss of weight, or after changes due to ageing, so structural invariances in mind and personality will also provide a means of identifying someone as being someone we know. The causal condition has already been used implicitly in what has so far been said about the parent-child case. For it is the mechanism of genetics that explains those structural similarities between parents and their children on which survival claims are to be based. Genetics thus gives us just the right kind of causal account. Incidentally, note that I am not defending any sort of extreme genetic determinism here. There is no suggestion that all a person's physical and behavioural characteristics are a parental legacy: the development of disease, the acquisition of agility and fitness, the possession of a sour or an optimistic personality, are things that may well depend on environmental and developmental factors. Indeed,

Identity and Continuity


my suggestion is the modest one that since some features of physical morphology and personality are genetically determined, theJl parents can make a claim to survive to the extent that such heriditary characteristics in fact appear in their offspring. Much bolder is the notion in T. H. White's children's novel, The Master, where the central character hopes to avoid wholly dying by kidnapping someone else's children and passing on to one of them his lore and values. 12 Here the Master's survival depends on the changes in behaviour and values that he can induce in others by teaching, indoctrination, and other such means. Although I think we can make sense of even this sort of case, and illuminate it by use of the three conditions, we would need to look carefully at it in terms of the account of mind and personality developed later. And what of the third condition, the one concerning matter? Thanks to the remarkable features of animal reproduction, we again see in the parent-child case a clear example of similarity of matter from parent to child, this similarity being explained by the mechanisms studied by genetic theory. As we will discover, the term 'matter', is not always clear: suffice it to say here that in terms of matter, understood at a fundamental biological level, parents and their offspring are constituted of just the same sort of matter-cells with certain structures, these depending in turn on the properties of a limited number of types of organic molecules. So our case of the parent-child survival can be described in terms of our three conditions, albeit recognizing that only some parental structures will reappear in any offspring, and that the reappearance of these structures and the fact that they utilize the same sort of material can be plausibly explained by reference to current biological theory. Notice, by the way, that although current biological theory gives good support to what is said here, it is by no means necessary to the 12 The Master's attempt to survive through other people unrelated to him reminds us that the transmission and preservation of culture and values may be just as important to survival as the (more currently fashionable) transmission of genes. At various places in the text I do depend on the naturalness of the idea that the passing on of genetic material is of significance, given certain conceptions of survival. It is worth noting explicitly at this point, however, that I would just as happily have used examples concerning the transmission of culture, political and moral values, religious or other myths, and so on. Complications about the kind of survival involved in these, and the genetic cases, will be explored later. In White's story, when the Master finds one of the 'detained' children is suitable for education, he says: 'Non omnis moriar' (l shall not wholly die). (T. H. White, The Master, London: Jonathan Cape, 1957.)


Identity and Continuity

account of survival. It is easy to imagine ancient Egyptians believing that they survived in their children; and superficial structural features and similarity of matter would underpin these claims. Anthropologists have reported, so I understand, that in some societies there is thought to be no causal connection between sexual intercourse and pregnancy. In such societies, if my proposals are roughly right, we would not expect fathers to believe they survived in their children. Alternatively, if they did hold such a belief, they would also hold some associated beliefs about appropriate causal relations between themselves and their mates' offspring (though these beliefs would be very different from our own). So nothing in what is proposed here ties the philosophical theory of survival to acceptance of particular scie.ntific theories. We can now turn to the other question we raised, namely, whether the account of survival can be applied to the survival of one stage of an object in or as some later stage of the same object. It should be clear that the prospects for doing this are quite good. Objects of various sorts change their material from time to time. Thus the mammalian pancreas has a complete change-over of cells roughly every twenty-four hours, yet in healthy organs this replacement is structure-preserving (and hence function-preserving). Small trees with thin stems grow into large trees with thick trunks; in this case we have structural as well as material changes. Again, though, we can look for the survival of one tree stage in some later stage: the geometry of leaf distribution shows some consistency over time, so that we can envisage the shape of the later stage by inspection of a stage of the immature tree. Likewise, later stages carry the historical evidence of pruning and damage suffered by earlier stages of the tree. In discussing plants and other such organisms, we have to note the biological distinction between the unitary and the modular. A unitary organism has a highly determinate structure, with the number and location of limbs and organs determinate (within the range of genetic variability). Human and other animals are clear examples of such organisms. By contrast, modular organisms are characterized by a post-zygotic phase in which a module of construction is developed which itself develops further modules, and so on. Modular organisms typically branch, and their growth and development depends strongly on their relationship to the environment. A tree will thus reveal not only the history of past

Identity and Continuity


pruning but, to the knowing eye, also the story of the mineral deposits in the surrounding soil and the direction of prevailing winds and available sunlight. As recent ecological work has noted, modular organisms show differential response to predator attack and invasion by disease organisms. Thus a tree behaves in some respects more like a community of modules than like a single particular. An extreme case of environmental influence on the development of a modular organism is that of bonsai cultivation, where perfect miniatures of trees and shrubs are grown in deliberately impoverished environments. Artefacts of certain sorts do not generally show such dramatic changes over their histories. A paintbrush tends to keep pretty well most of its material as long as it is able to function at all. Upon finding a brush that I had used when new, and which now bears the scars of a long working life, I may recognize both the constancies and the changes. If the bristles have clogged with paint, I may regret that its ability to hold paint and apply it smoothly has been lost; but it may still be a good width for painting certain awkward spots. Here I both recognize the survival of the earlier stage to some extent, and-at the same time---regret that it has not survived better! Of course, we normally just speak of this in terms of the properties of the brush itself. The point being laboured here is that we can readily translate into talking of survival of earlier stages inor as-later stages. In this chapter, identity has been distinguished from similarity. The formal principles of identity reveal it to be a congruence relation, that is, it is reflexive, symmetric, and transitive. Leibniz's Law (the indiscernibility of identicals) has generally been taken as less controversial than the converse principle of the identity of indiscernibles. Although it is natural to explicate the identity of a persisting thing in terms of spatio-temporal continuity, some theorists have preferred to appeal to indiscernibility as elucidating identity because of the existence of cases where more than one thing occupies portions of the same path through space-time. Continuity theorists deal with the same cases by explicating identity in terms of sortal-covered continuity. A relation of survival was then provisionally defined. Although suggested by the work of Parfit, the definition of the concept involves conditions concerning structure, matter, and cause which


Identity and Continuity

are not explicit in his treatment. The definition given can be taken as stipulative. Although the three conditions given appear to be sufficient for survival, only two of them-the cause and structure conditionsseem to be necessary. Since survival can hold to a degree, it is not surprising that sameness of structure and matter can also be a matter of degree. The nature of the causal condition requires more study, as does the scope ofthe structure condition. The programme for the immediately following chapters is the elucidation of survival and a consideration of its relation to identity. More consideration will also be given to the question of spatio-temporal continuity.

2 Metaphysical Problems 2.1. THE CHAIN THEORY

Philosophy is often concerned with theory-construction. Not all philosophers recognize this as a proper description of their work but if it is taken seriously, then my concern here is with theories of identity. We have already undertaken, in the first chapter, the sketch of a theory of survival, a theory which will have a significant role to play in the theory of identity. But to approach the identity or unity of a persisting object through the study of survival in the way implied in the first chapter is to take a certain view on what a unified object is. We can call this the chain theory of unified objects. On this conception, itself highly theoretical, a unified object, like a table, or cat, is a chain of temporal pans, all neatly linked to each other. The temporal parts, or stages, do not, of course, have the spatial property of interlocking: but we may expect them to be in some sense 'tied' in to each other. If an enduring, or persisting, object is like a chain of stages, then we can think of such stages as being links in the chain of its life. This immediately suggests that we should make a clear distinction between enduring things and their temporal pans. For although a chain is composed of links, no link on its own constitutes a chain. Rather, only a collection of interlocking links makes up a chain. Just so, we might think, a suitable collection of time slices 'glued' together according to our three conditions will constitute an enduring object. But the object is one thing of a kind; the temporal part is something of a different kind. Viewed in this way, the suggestions sketched in the previous chapter look like moves in a decent reductionist progranune. We are puzzled, let us suppose, by the notion of persisting thing: what is a unified spatio-temporally broad object? What is a body, to use Quine's term? The answer is that such a unified broad object is nothing more than a suitably connected sequence of object-stages (each of which survives as its successor); so a table is a suitably connected sequence of table-stages and a person is a suitably


Metaphysical Problems

connected sequence of person-stages. Since stages are links, bodies are chains, and links are of a different kind from chains, then we have explained the nature of one thing in terms of something else. Maybe we can even define a unified, persisting thing as a maximal Srelated sequence of object stages. Here, stages are S-related if and only if they are sequentially linked according to our three proposed conditions. However attractive this approach seems, it encounters several difficulties. One of these is that we have still to decide whether or not temporal parts of things are items of the same kind as the larger things supposedly constituted out of them. Links, as we have noted, are never chains. But it is not so clear that table-stages are never tables. After all, a table-stage is simply a table-at-a-time, and surely that is a table (at a time)! But if table-stages are tables, then our reduction of tables to sequences of connected table-stages is not so impressive. We will look in more detail at this worry later. For the moment, let us note a further appealing feature of the programme associated with the chain theory. Suppose that we use the term 'individual' in the mereological sense. The mereologist is someone who prefers, often for reasons of metaphysical austerity, to work with individuals instead of such 'abstract' objects as sets. In fact, the notion of abstract is just as bedevilling as its opposite, the notion of concrete. It is odd how many terms of art there are which can be readily acquired by students, used by professionals, dropped impressively into everyday conversations, and yet prove embarrassingly difficult to define or explicate. 'Concrete' and 'abstract' are good examples of just such terms. One way in which things can be concrete is by having effects or being acted on by causes. My desk and I are concrete by this account, for we causally act on one another (or, to be more precise, we are ingredients of events which are causally related to each other). My unit set and my desk's (the set whose sole element is the desk) are not causally interactive at all and thus abstract. Again, perhaps alternatively, to be concrete is to be 'in' space and time. Thus a concrete object may be something that is in a one-one correspondence with a set of points in space-time. As Kripke points out, such an identification would mean that no two objects could be in the same place at all times. On this account of the concrete, numbers are abstract for they, like sets, lie outside space and time.


Metaphysical Problems

Goodman's well-known account of mereology (the calculus of individuals) in The Structure of Appearance relies on the foHowing nominalist tenet: 'entities differ only if their content at least partially differs'. 1 Individuals are precisely what satisfy this tenet, but-it turns out-they can be 'abstract' in a certain sense. For our purposes, we tan start from this same conception of individuals, and let the worry about abstractness go. As we will see later, it is quite easy to slip into 'abstract' conceptions of objects without using any set-theory. The appealing feature of chain theory is the following. Suppose we think again about Theseus' ship (Chapter 1.4). If stages are individuals, collections or fusions of stages are individuals as well. This follows from a happy feature of the mereology: if x andy are individuals, however disparate, then x together with y is also an individual. So suppose we represent the ship case as shown in Fig. 5. Here we take A, B, and C to be names of collections of individual stages. This is a departure' from our practice in the previous chapter where the name 'Heraclitus', for example, designated the broad continuously functioning ship consisting of A + B in the first two situations. Since each stage is an individual, then a collection of such stages will also be an individual. And, of course, A + B, A + C and so on, will also be individuals. So in our three diagrams, how many individuals are represented? The answer to this question is quite unclear. For we do not know how


B .................... "






Fig. 5.

I See Goodman [1], p.36. Goodman tries to clarify his understanding of nominalism in the following passage: 'If no two distinct entiries wharsoever have the same content, then a class (e.g. that of the counties of Utah) is different neither from the single individual (the whole state of Utah) that exactly contains its members nor from any other class (e.g. that of acres of Utah) whose members exactly exhaust this same whole. The platonist may distinguish these entities by venturing into a new dimension of Pure Form, but the nominalist recognises no distinction of entities without a distinction of content.'


Metaphysical Problems

many stages are in each of A, B, and C. But we can answer the question: how many individuals constructed out of A, B, and Care represented in the diagrams? For the answer to this is: three-A+B, A+C, and B+C. Now, although each ofthese collections or fusions of A, B, and C are individuals, it does not follow that each is a unified enduring object, let alone a persisting ship. Take B + C, for example. They coexist in the situation depicted in the middle diagram, and constitute together one individual, but this individual is not something that we would recognize as itself constituting anyone thing of a kind or sort. Likewise, my desk-top and a rat in Bali constitute one individual, but we lack any generic title for such a combination. Thus, there are more individuals-a great many more--than there are unified objects of sorts we recognize. The chain theory therefore confronts the following interesting problem: since we can construct any number of chains from the links available, what features make certain of these chains stand out, so to speak, as unified, persisting objects of a sort? The problem here is no more (and no less!) than the temporal analogue of a similar problem concerning spatial parts. Some collections of spatial parts stand out as unified objects, as we have seen, and some do not. The proposal in Chapter 1 was to use the idea of survival in the account of chain construction. The crude idea, applied to Theseus' ship, would be that those stages belonging to B-indeed which constitute B-have suitable survival relations to each other. Each stage in B survives as its neighbour in the chain oflinks constituting the large individual, B. But no stage of B survives as a stage of C. So we have no grounds, based on survival, for linking the stages of B together with those of C into a unified individual B + C. By contrast, any stage of A survives as some stage of B and also as some stage of C, in the cases where C is present. We can construct plausible sequences of stages, then, which would allow us to count each of A + B and A + C as unified objects. A worrying question, which we defer for the moment, concerns the conditions under which we can allocate such unified objects to kinds or sorts: in the situation where both Band C coexist, it has seemed to some that A + C cannot be a ship, even if all the stages of A and Care themselves ship-stages. Leaving this tricky matter aside for the moment, we can ask whether, from the point of view of chain-builders, there is any alternative account that would supply the unification of A and B in

Metaphysical Problems


the leftmost case of our diagram, of A and C in the rightmost, and yet make no appeal to survival. One standard account does attempt to do this-the account in terms of spatio-temporal continuity. It is a virtue of the chain theory that it allows us to formulate the question of unity in terms of stages or temporal parts: what properties does a collection of stages have to possess in order to constitute the links in the history of one unified thing? The continuity theorist is able to give an apparently straightforward answer to this question in terms of a sortal-covered, continuous path through space time. It is not clear, however, why an enduring or persisting item should not be episodic, that is, have gaps in its spatio-temporal history. This possibility is to be explored in the fourth chapter. However, it will become clear from simple examples to be discussed in the third section of this chapter that continuity is not an easily defined notion. Worse, the problem of Kripke's disc-to be discussed in the fourth section of the chapter-shows that the links, conceived as momentary things, will not give us the basis for constructing an account of moving bodies. Kripke's work on identity forces us to confront the issue of primitiveness, an issue that will recur in various forms later in the book. Although objects or particulars are distinct from their trajectories or histories in space time, this does not mean that the unity of particulars is ultimate or primitive. Particulars are not restricted to three dimensions; as will be argued in the fifth chapter, there are ways of thinking about particulars as three-, four-; or even five-dimensional. In the present chapter, I will be trying to show that in giving an account of the unity of one sort of thing, we are able to take the unity of other sorts of things as primitive; primitiveness is relative to the explanatory exercise. More objections to continuity theories will come in the fourth chapter. For three-dimensional theorists, as I willsuggest later, it is hard to phrase the unity question at all, while ignoring the distinction between metaphysics and epistemology would simply make the unity issue even more confused.


We have already seen (in Chapter 1.2) how some theorists take the notion of a persisting thing to be a notion of an item which traces a


Metaphysical Problems

continuous path through space and time. And at first sight, the notion of a continuous path of this kind seems useful, provided a certain number of problems can be overcome. In Chapter 1.3 we saw how continuity and indiscernibility theorists could come to different views on the Rover/Clover case. This time, of course, we are taking the continuity theorist as engaged in tracing the history of a broad object by following the history of successive and continuous stages. One alternative suggestion for dealing with the Rover/Clover case is that the path starting with a Rover-stage, although continuous with one ending with a Clover-stage, is not one that can be traced throughout under the same sortal. Although we mentioned this manreuvre earlier, we did not then pursue in detail the difficulties with it. The most profound of these is not whether a sortal still applies to an item when, for example, it is dismantled. There seems no problem about pointing to a bundle of nylon fabric and a pile of aluminium poles and saying 'That's our tent.' Many artefacts are designed for regular dismantling-gliders, sailing dinghies, inflatable boats, tents, many kinds of gun, and so on. Even those that are not meant for leading such functionally discontinuous lives, can be taken apart for repair and reassembled. Thus an automobile engine can still be traced under a sortal while scattered over the garage floor. The dismantling, we may say, prevents its functioning as an engine, without thereby leading to its ceasing to exist as an engine. Some theorists-Hirsch and Quinton, for example-take functional discontinuity as a case of going out of existence. 2 For them, the disassembled works scattered over the garage floor do not constitute an engine at all. Rather the engine goes out of existence when dismantled and comes back into existence when reassembled (provided enough of the same components go back in without being replaced). Now this difference of perspective is nothing to get hot under the collar about. Very little, other than tenninological conventions, depends on whether we adopt my way of speaking about functional discontinuity, or their way of speaking about lapses of existence. The important thing for a continuity theorist is that, even if the engine is said to go out of existence when dismantled, there is spatio-temporal continuity of enough of its 2

See Hirsch [2], ch. 1, and Quinton [2], pp. 63 fr.

Metaphysical Problems


parts to sustain the claim of identity upon reassembly. We can call this the requirement of compositional continuity, and let us suppose for the time being that we can come up with some reasonable standard for this. So a continuity theorist can take two lines on this question of dismantling. One is to treat the space-time path as still continuous and still the path of that object (this is the advantage that comes of talking about functional discontinuity). For this kind of view, there is a separate issue to confront, of course, in the case of dismantling, as also in the case of things which can survive renewal of their parts. This is the quite general question of how tolerant an item is to replacement of its parts, and since it is an involved question, we defer it for the moment. The second line the continuity theorist can take in the present case is to recognize discontinuities in the existence of objects under certain descriptions, but supplement the continuity story in such cases by appealing to continuity of (enough of) their parts. Clearly either line might still permit us to think of a continuous path traced though space-time by objects and their parts. So where are the more profound difficulties with the continuity view? It looks as if it has a certain advantage over the survival account. For it is surely simpler to regard the Theseus' ship case in terms of continuities among the stages of A and B and of A and C (but not of B and Cl) than in terms of our more cumbersome conditions on survival. The two big problems for this account arise in connection with the phrase 'sortal-covered continuity'. The first problem is to say what it is for a stage of an object to fall under a sortal. And, secondly, we have to ask if we can give a satisfactory account of spatio-temporal continuity which does justice to what we usually count as continuities in an object's history. The first big problem has already been mentioned earlier in the chapter and will be mentioned in subsequent chapters before we get round to discussing it properly. Even then, in our discussion of possible worlds and possible things, we will end up with only a tentative conclusion. The second problem will be dealt with now. In subsequent chapters we will be exploring the survival account in more detail, and so an explanation is owed at this point to those who think of spatio-temporal continuity as a clear and helpful adjunct to the theory of identity.



If, like David Wiggins, we take identity as a primitive notion, we might wonder why we should talk about it in terms of indiscemibility, spatio-temporal continuity, survival, or anything else. My own view here is that we take the identity of some items as primitive in our theory construction; in terms of these primitives we can give an account of the identity of other items in any terms that seem appropriate. Thus, for example, our account of the identity of a table in terms of certain relations among table-stages, presupposes that we have an understanding of what it is to be one and the same table-stage. But the account gives us no way of explaining the presupposed identity. For this reason, as several theorists have noted, it would be better if we dropped the term 'identity' from treatments of this topic. 3 Instead, we might talk about the problem of the unity of tables, lakes, engines, or whatever. It is important to recognize this problem about primitiveness, for it is a source of confusion to many people. If identity is always primitive, then of course there is little point in a theory of identity for any sort of thing. Wiggins emphasizes that his notion of the primitiveness of identity 'in no way excludes discursive elucidation in collateral terms,.4 And one of the ways he suggests giving such elucidation is through recognition of the fact that what organizes our method of determining identity is the idea of a continuous path 3 Compare Perry's remarks in Perry [7], for example. He writes: 'Let us say that stages of a single human being are H-related .... Our speculation is that ... Hrelated stages are, at least for the most part, related by some other relation (or at least by a relation that, for all we know, may not be the H-relation), and their being related by that other relation explains the effect of the earlier stages on the later. This new relation ... is my candidate for the analysis of personal identity. That is, it is the unity relation for persons ... ' Even if dropping the tenn 'identity' is going too far, it is certainly valuable to reconstrue thoughts about identity as thoughts about unity. 4 Wiggins [9], p. 49. If we take identity as a congruence relation, namely that relation holding between a thing and itself, it becomes hard, as Frege pointed out, to distinguish between· interesting identity statements, of the form 'a = b', which can add to our knowledge, and trivial statements of the fonn 'a=a'. For, where 'a=b' is true, it says that a is the same thing as itself. This led Frege to a semantic account of how the two kinds of statements differ; see 'On Sense and Reference' in Geach and Black [1]. For discussion of the interpretation of identity statements, see Morris [1]. The semantics of identity statements is not of particular importance to the present study. For those who take identity as being primitive, there is no more to be said about the unity of particular things except that they are unitary. The semantics of identity statements will, for them, be the natural focus of study in this area.

Metaphysical Problems


in space and time. We will return to the issue of primitiveness at the end of the chapter, so let me concentrate just now on the 'elucidation' in terms of continuity. One point that Wiggins makes clear is that any notion of continuity or gappiness we use must be appropriate to the kinds of object we are studying. We have seen already that some things are expected to lead functionally discontinuous lives. But we need to be sure at this point that, in other cases, we do not confuse continuity with some other relation between objects or their stages. There is a Hollywood sense of 'continuity' which has a certain vogue in these discussions, and we must be wary of confusing this with spatiotemporal continuity. In the Hollywood sense, two sequences in a film show continuity provided a number of conditions are met. These include similarity of clothing and make-up worn by those on screen, similarity oflighting conditions (so that a sequence in which a car stops in broad daylight is not followed by one in which the occupants get out in pitch darkness) and so on. Observance of strict continuity standards is essential to fooling the audience. For it ensures that the audience views sequences that have been shot separately and often non-consecutively as if they form one unbroken stretch of action. Robert Nozick has introduced the idea of a close continuer into discussions of identity, where this is something like the Hollywood idea. On Nozick's account: to be something later is to be its closest continuer.... To say that something is a continuer of x is not merely to say its properties are qualitatively the same as x's, or resemble them. Rather it is to say they grow out of x's properties, are causally produced by them, are to be explained by x's earlier having had its properties, and so forth.5

Nozick illustrates what he means by an ingenious experiment using a screen. Suppose we view the situation in Fig. 6. If x disappears behind the screen moving right, and y comes out appropriately later, at a slightly different angle but looking like x (same shape, colour, and so on) and moving at the same velocity x had, then we will, according to Nozick, see y as the same object as x, deflected by collision with something behind the screen. However, if-in competition withy-an item z emerges, again with a suitable 5

Nozick [11, pp. 33-5.


Metaphysical Problems


__ x

Fig. 6.

delay, and looking and moving just like x-then we will tend to identify z with x if the track of z is closer to x's original track. In this case, y is a close continuer of x, but z is a closer continuer. Here we have a nice vocabulary for applying to the Theseus' ship case. For our problem there is how to pick out the closest continuer: the closest continuer of the original ship, A, will deserve our vote for being Theseus' ship. The trouble here is not so much Nozick's approach to the unity question, but his choice of vocabulary. For, of course, in the screen experiment, there need be no spatio-temporal continuity between x and either y or z. For all an observer with limited information might know, x just stops dead behind the screen and some completely new object is launched on an appropriate trajectory. The continuity here, then, might very well be of the merely Hollywood sort. We will look in Chapter 4 at more cases like this, where identity is discussed in relation to episodic objects. The important lesson we should learn at this stage is that cases of unity may be envisaged where a single object can be supposed to have gaps in its spatio-temporal history. But then such objects would not be things that trace a continuous path in space-time. So what, then, is a continuous path through space and time? This question is rather harder to answer than we might at first have thought. For we seem to have, on the one hand, a clear notion of what such continuity is; but when we consider even mathematically naive formulations of the notion, we run into severe difficulties. If, following Hirsch, we define the place of an object at some time as the region of space coinciding with the object at that time, we can say that places overlap if they have a part in common. Now one notion of continuous spatio-temporal change is that if we take successive times t and t' at both of which an object exists, then we can make the extent of overlap between the places of the object at t and at t' as great as we like by taking t' suitably close to t. 6 Negatively, we can 6

Hirsch [2], pp. 15-21.

Metaphysical Problems


put the same point in the way Kripke does it in his lectures on identity. If we take two successive volumes of space occupied by an item then we can make the extent of non-overlap of the two volumes arbitrarily small if we make the times arbitrarily close. One big problem with this simple version of the mathematical notion of continuity is that it fails to do justice to those continuities we recognize in daily life. Hirsch shows this by taking the example of a tree. Suppose we lop a large branch off the tree. Now, although, as we cut, the branch may slowly move downwards, it is still attached to the tree until, with the last stroke of the saw it is completely severed. But then there must be an instantaneous change in the tree from the' volume it occupied when the branch was attached to the volume it occupies once the branch has gone. In other words, the continuity of the tree is compatible with a real discontinuity in its volume from one moment to the next. Hirsch suggests one way we could try to patch things up here. Maybe the notion of continuity we want is one like this. Take any two successive volumes occupied by an object. Then we can still maintain a moderate degree of continuity if we let the extent of overlap between the volumes be greater than the extent of non-overlap.7 Thus, we cannot instantaneously deprive a tree of branches accounting for more than half its volume, on this view, without thereby destroying the tree. Obviously, we could go for even weaker kinds of continuity. Maybe we could permit objects to have discontinuous changes provided there is always some overlap between successive volumes occupied by the objects. It is clear that we would need to take the weakest view of continuity if we were to allow that items like rockets can stay the same while shedding their various stages. Consider, for example, exactly what happens when a multi-stage rocket jettisons its first stage. This large piece of equipment, being mainly a massive fuel tank with motors attached, may well account for an enormous share of the total volume of space occupied by our original rocket. its instantaneous jettisoning thus accounts for a loss of volume that prevents us saying the history of the rocket is any more than weakly continuous. However, we cannot rest content with weak continuity in Hirsch's sense as a general account of what we mean by saying that 7

Hirsch [2], pp. 10--14.


Metaphysical Problems

objects trace continuous paths through space and time. As Kripke points out in his lectures, if we were to watch the motion of a rolling ball that was merely weakly continuous then we would see not continuous movement but a series of jumps or jerks in its behaviour which-however appropriate to subatomic particles-is hardly what we expect from macroscopic objects like balls. Just as Locke remarked that identity is suited to the idea, so we might want to say that standards of continuity have to be made appropriate to the objects under investigation. But any such admission seems to threaten the simple view that the notion of a continuous path through space and time is intuitively clear and mathematically definable. As if this were not a big enough problem, our attachment to continuity is liable to be threatened by a further one. We could take a stab at it by asking: what is it we are judging continuity of? Here I am, looking at a tree. And over there is the tree. It doesn't move, so it shows more than moderate continuity from moment to moment. Even if it sways gently in the wind, it still shows more than moderate continuity. For the closer I take the successive volumes it occupies, the greater the overlap between them. But there are lots of other continuities here too, although we are not usually inclined to recognize them. For example, there is a high degree of continuity between the tree just now and the tree a few seconds later when trimmed of a few millimetres all round. And between that diminished tree and the tree diminished by a further few millimetres all round there is also continuity. Of course, we don't normally recognize the existence of such continuities: but those individuals are all there even if we don't see them. For the continuous path of the undiminished tree there are thousands of alternative paths of diminishing and growing objects, also highly continuous. So why do we pick the one we do? Hirsch's tentative solution to this problem is that we trace the tree as a continuous object and ignore all the alternative possibilities because to do so minimizes change. Indeed, he takes sortal-covered continuity to be a refinement of this more primitive notion of change-minimizing continuity. And this looks plausible. For certainly a Papuan who had never before seen a car could no doubt trace the history of a car for a certain time as the history of one thing even though lacking the appropriate sortal vocabulary and the associated concept. But this is no more than a plausible suggestion.

Metaphysical Problems


We are . left with the feeling that we cannot talk about change minimizing continuity unless we know just what our frame of reference is. Relative to what, in other words, is change to be minimized? Related to this problem is the epistemological question of how anyone could start from the links, taken as momentary slices of objects, and build up from them continuous and continuously changing broader objects. Kripke has focused this issue for us by discussing the problem of a rotating disc, and to that we now turn. s


It is helpful, in coming to understand this problem, if we follow Kripke in thinking about what I have called the 'chain theory' in a slightly different light. Recall that the chain theorist is trying to give an account of broad objects in terms of connections among their temporal parts. At the moment, we have two accounts of how this is to be done. On the survival account, whole objects are constituted of temponil parts---object-stages-suitably linked according to our three conditions on survival. On the continuity account, the links between stages consist of appropriate spatio-temporal overlap of the stages, the degree of overlap standardly depending on the nearness in time of the stages in question. Kripke suggests that we think about the links in our chains as follows. Each link is part of an enormous, three-dimensional 8 Kripke's lectures are unpublished, but widely circulated in manuscript form. The disc example is just one of several that pose a problem about spatio-temporal continuity. For a detailed discussion of it and for further references, see the article 'Identity, Properties and Causality' in Shoemaker [6]. In the same article, Shoemaker provides examples that cast doubt on whether spatio-temporal continuity is even sufficient for identity. I had long ago tried to argue for the same conclusion as Shoemaker, using the example of a projected picture. If we could agree that the identity of the projected picture was determined by the identity of the slide in the projector, then continuity of projected picture would be compatible with difference of picrure (for projection of one slide can be stopped at the same time as projection 'of a different slide was commenced). Although Shoemaker's example is more fanciful than this one, it is perhaps more convincing. He imagines that we have machines that bring tables into existence and which can also cause them to dematerialize. But, given such machines, a continuous series of table-stages may well be stages of different tables-for a given table is made to materialize by the tableproducing machine at just the very time and place that another table was destroyed by the table cancelling machine.


Metaphysical Problems

photograph of the world at a time. We could call such a freeze frame a hologram of the universe at that moment. The important thing is that we be able to read off from the hologram the complete state of the world at the given time without prejudice to whether successive holograms are showing the same or different objects. An analogy with experience may help here. Clearly, I can have two experiences separated by, say, ten seconds. I look at a cat in the corner of the room, close my eyes for ten seconds, and then look again. Clearly, my visual experience when I look again can be so described as not to prejudge the question of whether I am seeing the same cat again or simple a different, but extremely similar, cat. The holograms imagined by Kripke are likewise to be thought of as having all the information we need about the state of any items in the world, but not information about their numerical sameness with items depicted in earlier or later holograms. Now we imagine a thinnish, circular disc about which we have just such information. In hologram after hologram we glean the information that the disc is apparently sitting there in the same location. So let us try to track the disc's history. One way is to say that the disc just sits there. Each time we inspect the hologram, we put a blob on part of the disc, or shine a light, or whatever. So in Fig. 7 is one tracing of the disc's history through time: the spot in each case is meant to mark the same part of the disc each time. But equally continuous is the track of the disc's history seen in Fig. 8. We can ensure that this mode of tracing is continuous according to our earlier accounts: if we shorten the time between successive holograms then we move the spots closer together. Moreover, each

000000 000000 Fig. 7.

Fig. 8.

Metaphysical Problems


time, we are not only tracing continuously but tracing under the same sortal (provided we can apply sortals to instantaneous stages). So, how can we decide-at least on continuity grounds-between these two modes of tracing the disc? One tempting reply to this question does no good at all. According to it, the answer is to be found in Hirsch's notion of minimizing change. For surely, of the two proposed accounts, the former is to be preferred simply because it does minimize change. But this won't do at all. The case of the disc is significantly different from the case of the tree where minimizing change looked like a good idea. For we have no information, so far, as to whether our disc is rotating or not. Our successive pictures will be the same from instant to instant whether the disc is rotating or at rest! Here is the real trouble with the freeze-frame approach to the description of the world: it just cannot provide us with the information we need in order to know if the disc is rotating or not. Kripke points out that it must follow from this example and similar cases that the hologram view of the world is completely flawed, certainly so if we thought that from a sequence of such holograms we could deduce facts about the identity or otherwise of objects. Nor is the deduction theory saved by attempting to construct suitable counter-factuals. We might say, for example, that the disc is rotating if, were we to paint a white spot on it, the spot would show up in different positions in successive holograms. This would hardly save the day. For surely we cannot define rotation by this means. The changing position of the spot would, of course, be evidence of rotation, but the rotation would be the independently specificable fact that explains the movement of the spot. What conclusions can we draw from this case? It might be thought that a Quinean move is attractive here. As we have seen, Quine distinguishes bodies (unified, broad objects) from physical objects (which are any old mereological individuals we care to take). Now, for the Quine of Word and Object, bodies are posited by us; or, more accurately, we learn the body theory as we grow up because bodies were posited by our ancestors in an inspired bout of theorizing. 9 Theorizing-in Quine's view-is not to be taken 9 See Quine [2], § 6. Quine's view on posits and reality is modified in his later work: see Quine [9], where, as noted in the text, he comes round to a position closer to Hirsch's on body-mindedness (see Hirsch [2], ch. 8).


Metaphysical Problems

lightly. 'To call a posit a posit,' he cautions us, 'is not to patronise it.' So we could try defending the hologram model of the universe from Kripke's attack by claiming that the features and unities of bodies are postulated on the basis of the freeze-frames as data. In the case of the rotating disc, the data permit the generation of alternative theories. And in this case we are as likely to come to the wrong theory as to the right one. Luckily, though, the universe is not stocked entirely with uniform symmetric objects for which the rotation puzzle arises. So in most cases, we come to theories about bodies that are not in competition with close rivals. The idea that our body-mindedness is the result of this kind of theorizing seems an attractive response to Hume's problem about the continued and independent existence of external things. Our access to the world is, clearly, partial and incomplete. Empiricist theories of knowledge and perception have often used the notion of inference in an inappropriate way. If we are to infer the existence of bodies from our experience then it has seemed that a dilemma must be faced. Either there is more to the nature of bodies than is given to us in experience--in which case the inference cannot be logically or deductively justified--or bodies are themselves in some way constructions out of experience--in the way phenomenalists, for example, have claimed objects to be logical constructions out of sense-data. But in the second case, the empiricist seems driven into a kind of idealism--of the sort found in Berkeley. The theory of knowledge is not, of course, a central concern of this study. But we cannot dwell long on problems about identity without these leading us into other central areas of philosophy. We will not be able to solve some of the problems we face without further excursions into the theory of knowledge. Hirsch, interestingly, finds himself facing the problem of body-mindedness too. For him, our 'sense of unity', as he calls it, is itself an innate endowment. I think there is quite a gap between Hirsch and other empiricists here, although Quine has come round to thinking in terms of body-mindedness. Even on Quine's earlier views, the theory of bodies needs some help from our innate abilities. However, what Quine thinks is innate in us is a kind of biased spacing of qualities, an ability to treat objectively similar situations as in fact dissimilar. In our later discussion of objective similarities we will see how Quine has solved a problem that some would dismiss as unreal. For it requires us to take seriously the existence of

Metaphysical Problems


objects as mereological individuals,. a requirement that can be dismissed by opponents of what in Chapter 7 I call the extensionalist myth. Returning to Kripke, we can ask if the suggestion that we have a theory of bodies solves the immediate difficulty. Unfortunately, it doesn't. For what Kripke seems to be arguing is that the motion of bodies deserves to be among the basic facts from which any further theories are generated. Why should we rest content with an entirely artificial picture of how we come to our views about unity? If I cannot tell, in a given case, whether a disc is rotating or not, I can always put out my hand and touch it. Of course, if! am looking at an expensive piece of equipment, or a dangerous one, there may be good reasons for not pokirig around in the works. But, in many cases, such an option is available to me: even if I can't decide by looking, I can decide by feeling, whether the disc is rotating or not. If we insist on restricting ourselves to instantaneous cross-sections of the universe, we cannot include motion among the data. And Kripke's arguments are fatal to the suggestion that motion can be inferred from instantaneous descriptions. Of course, Kripke's disc example is not fatal to the idea that enduring bodies are theoretical posits. So we have to be clear that, where we are worried about conclusive arguments, his example does not seem to mark the end of bodies as posits. But we can perhaps appeal to good sense here. Quine is right to suggest that we can draw the dividing line between theory and data virtually anywhere we like. But it would be perverse, unless initiated by very compelling reasons, to draw it in such a place as to rule out motion in objects as part of the data used in constructing theories of the world. So let us agree with Kripke that we can identify things as displaying changes of motion without having to infer, or theorize, this motion on the basis of instantaneous data. This granted, though, it does not seem to follow that the concept of an enduring or persisting thing is primitive and irreducible. Motion is a form of change which affects an object. Kripke seems to think that if change of this sort is admitted as a datum, then we have to accept the primitiveness of broad, enduring things. But how long-lived need something be for us to detect its motion? Of course, if we look at a freeze-frame photograph, we will not detect motion, either of the camera or of the object, provided our shutter-speed was fast enough. But at shutter-speeds of one-thirtieth of a second it is


Metaphysical Problems

usually easy to tell if the camera, or object being photographed, was in motion or not. At exposures of one or two seconds, it is yet easier to give such verdicts. So we have to note that Kripke's motion example does not force us into accepting as primitive objects that are very broad. Of course, for some rare elements, a lifetime of one or two seconds would be long indeed. But in the history of people, rivers, mountains, and planets, a slice of one or two seconds is a miniscule portion. It would be odd, in fact, to approach the question of object-stages with instantaneous stages in mind-although we might have to bring instantaneous stages into our story in order, for example, to give an account of continuity. But, as we have already seen earlier in this chapter, there is no general, intuitively compelling account of spatio-temporal continuity which matches the changes in persisting objects. And, as we will see in Chapter 4, it is relatively easy to imagine a world (perhaps one with different physical laws from this one) in which a sequence of genuinely discontinuous stages can all be allocated to one object. That example, together with the comments in this chapter, establish fairly conclusively, I think, that a persisting thing need not have a spatio-temporally continuous history. And this is an important result, for it may be that human beings, and other animals, are precisely items of a sort whose mental histories are discontinuous in this way. In mounting the defence of the chain theory against Kripke's arguments, I have accepted his example, without querying its plausibility. However, it is clear that some difficult issues would need to be resolved before we can rest content with the example as given in his lectures. Suppose the disc of the example is not isolated from other physical systems, but, like the turntable of a record player, is connected causally to surrounding objects. In this case, although our holograms would not reveal obvious changes in the disc itself from moment to moment, they would reveal changes in motors, drive belts, speed-controlling devices, and the like, which are linked mechanically to it. Remember that the holograms give us as much information as we want to have about the state of the universe at the time in question. From all this information, it would be possible to infer that the disc is indeed rotating. Alternatively, think-in case this is what Kripke has in mind-of the disc as constituting a relatively isolated physical system, alone in deep space. Now we have the problem of specifying the frame of

Metaphysical Problems


reference relative to which we are ~o make sense of the disc's rotation, and of doing so in a way that leaves undecided the fact of its actual motion. I have not explored these kinds of problems here because I am happy to keep epistemology distinct, for the time being, from metaphysics. The metaphysical significance of the chain theory is that it gives an account of what it is for an item to be a single, unified, persisting thing. From such an account of the nature of things, nothing follows immediately about how we come to know their properties. So, a defender of the chain theory can accept Kripke's point about the impossibility of deducing motion from freeze-frame photographs. Further, even conceding that the individuation of a stage involves essential reference to the object whose stage it is does not threaten the theory that persisting things are chains of suitably linked stages. Not until Chapter 7 will the issue of the links between metaphysics and epistemology be explored. In that chapter, I will attempt to forge at least one link between the two domains by proposing some tentative explanations for how we come to recognize the world as containing items belonging to sorts.


We have come up with a number of suggestions in this chapter. One is that continuity in space and time is not as clear a notion as we might have expected it to be; another is that the concept of enduring things which change may be as primitive as any other concept used in constructing our theory of the world. Further, it has now been suggested that enduring things may themselves enjoy discontinuities in their histories. Yet we have stopped short of saying that everyday broad objects-tables, trees, dogs, or human beings-are primitives in our scheme of things. Instead, we have admitted that the slices or stages of such things with which we start will typically be at least of several seconds' duration, and it may be that stages spreading through several years are just as significant for our theorizing. Those who are familiar with Kripke's lectures may still not know to what extent I share his views on the primitiveness of identity. Therefore, some clarification of the primitiveness issue will be undertaken in this section. For Kripke, the identity of broad objects


Metaphysical Problems

is primitive and irreducible, while it is clear that on my approach there is a theory of their unity to be given. But this disagreement may not be as great as it appears. Moreover, it is important to recognize what is right and what is wrong in the widespread belief that the identity of persisting things is itself something that is ultimately fundamental. On the wrong reading of this belief, it is hostile to the entire project of this work; while on its correct . reading, it simply records a feature of experience. We have to take care that we do not get into a position that is hostile to explanatory and theoretical endeavours. Desks are solid, and support books, lamps, elbows, and sandwiches. Diamonds are hard, and make impressive scratches on glass and other materials. Yet-although both of these could be said to be pretty basic facts of experience-we should not rule out the possibility of explaining the solidity of desks or the hardness of diamond. Indeed, both of these things are perfectly well explicable given even a slight acquaintance with the science of materials. So I am happy to admit that persisting things, like solid and hard things, are fundamental ingredients of our experience. But if we are not debarred from explaining hardness or solidity by appeal to the interrelations of the components, or parts, of things, why should we not attempt to explain the persistence of things by reference to the interelations of their parts? In the latter case, of course, the parts in question will be temporal, not spatial, parts. In general, those who are doubtful of the enterprise of explaining four-dimensional persistence are so because of qualms they have about temporal parts and the associated chain theory of objects. Let me now look at a number of worries that are commonly expressed, and give some reassurance to those who may require it. First, a point made by Kripke in his lectures is the worry that any four-dimensional view of objects may rule out the possibility that two different things can occupy the same place and the same times. Think of Hirsch's example of a tree and its trunk. 10 Although trees are not trunks, if we forget about the roots and think of a tree with all its branches removed, we can envisage a case where-for some considerable time-the tree and the trunk may occupy the same places and times. Likewise, in Gibbard's well-known case of a 10 'When the space-time paths associated with objects partly coincide and partly diverge we have a case in which two (strictly) different objects occupy the same place at one time .. .' (Hirsch [2], p. 61.) cf. Lowe [2].

Metaphysical Problems


statue formed from clay, we can think of a lump of clay which comes into existence at the same time as the statue it constitutes and which ceases to exist at the same time as the statue. 11 Surely the statue and the lump of clay are two distinct things. Yet, looked at from a mereological, four-dimensional viewpoint, the tree and the trunk are, for a time, the same sum of stages; likewise, the statue and the lump of clay are the same sum of stages and thus indistinguishable. This concern is not unique to four-dimensional approaches. Anyone who thinks that a statue-or a lump of clay-is wholly present at a time, but who then goes on to describe the statue as being simply an aggregate of clay particles, is faced with a similar difficulty. For precisely the same aggregate of clay particles constitutes the statue and the lump of clay. One response to this difficulty is to maintain that statues, and indeed lumps of clay, are distinct from the aggregates of particles that constitute them. Mter all, if we smash the statue, we will still have an aggregate of clay particles (where 'particle' is not being used to refer to any fundamental portion of matter). Likewise trees, and tree trunks, are distinct from the aggregate of cells that constitue them. Applied four-dimensionally, this solution would mean distinguishing objects from their life histories, or trajectories, in space-time. As we will see in Chapter 5, there are very good reasons indeed for making just such a distinction. A unified object, or persisting thing-what I will be officially calling a particular-is not, on my account, bound either to three or to four dimensions. So I agree with those who maintain the distinction of objects from their life histories, but I do this for reasons that are quite different from those generally given. Hugh MelIor, for example, writes: 'a thing's trajectory, like Churchill's life, can itself only be identified by appeal to the persisting identity of the thing whose trajectory it is. Not every timelike path in space-time, whose parts are all occupied by events, is the trajectory of a thing.,12 Although I agree that we 11 See Gibbard [1). Noonan has recently defended the claim that in the case described, the statue and the lump of clay are one and the same. This predilection for concreteness on his pan forces him into a strange position. Since the predicate 'might have been squee2ed into a ball and not destroyed' is true of the lump of clay but false of the statue, this-like other modal predicates-must signify a different propeny when concatenated with sIngular terms differing in sense (see Noonan [8]). Luckily, I am not wedded to concreteness in this way; as is shown in Ch. 5, persisting things are not to be identified with their actual paths through space-time. 12 Melior [1), p. 286. Melior's paper is a critical response to Nerlich [l).


Metaphysical Problems

need to distinguish things from their trajectories, and also agree with one reading of Mellor's claim about identification, this does not force me to give up the attempt to explain the unity of a thing by reference to the relations among its temporal parts. Moreover, considering the spatial analogue of my temporal claims will give further support to them. Consider, then, the spatial analogue of Mellor's identification principle in the above remark. We can think of the spatial extent of an item as being similar to its temporal extent. Not every region of space whose parts are all occupied by things is itself a region occupied by a unitary thing. Because we know the typical spatial e.xtent of things like bicycles, dogs, and bridges we are usually able to identify the spatial extent of something once we know what kind of thing we are looking at. But now consider an attempt to give an account of the spatial unity of a desk along the following lines. We claim that the desk consists of certain parts, or matter, placed in certain structural relations and with certain kinds of causal dependence exhibited among them. The causal story will involved showing that, for example, the top and the sides, stand in certain relations of support to each other that do not hold between the components of the desk and other things. Moreover, if we apply a suitable force to one side of the desk, the other side will move. Changes affecting one part of the desk thus have consequences on other parts. We thus explain the independent movability of the desk and thus why it has an extent which is the extent of a thing, while the desk and the floor-although entering into some causal relationsdo not constitute one spatially unified thing. Now readers may make up their own minds whether they think there is any merit in such an approach to the question of spatial unity. The moral to observe here is that, even though spatially unified items are pretty fundamental in our experience, there is no reason why an account of spatial unity along the above lines should not be ventured. Moreover, if I were to propose such an account, I could do so while maintaining that desks are distinct from the sum of their parts and while admitting that identifying a certain portion of space as being occupied by a sum of desk parts requires me to have the concept of a desk. Likewise, my account of the temporal unity of an object is not to be seen as denying any epistemological claims about concept exercise; nor should I be accused of failing to distinguish items from their histories or trajectories. As already mentioned, there are very good reasons for dis-

Metaphysical Problems


tinguishing objects from either their three- or four-dimensional extents. Even those, like MelIor, who think that objects are in some way more fundamental than events, are prone to think that to talk about an object, or a persisting thing, is to talk, essentially, of something three-dimensional. On my account, however, there is no good reason for taking up this stance-and rather good grounds against it. The fundamental grounds for adopting the view that objects can be regarded as three-, or four-, or even as fivedimensional will be given in Chapter 5, and I will save most of my positive suggestions until then. In the meantime, largely negative arguments will be used to wean away those who are devoted to the three-dimensional conception of objects. A second objection to a four-dimensional treatment of persistence is related to the first. Since an object, conceived fourdimensionally, is really just a long event (or a chain of events), the four-dimensional treatment threatens the very distinction between object and event. Yet surely, the objection goes, this is wrong. Objects are what participate in events, or changes. A world of unchanging objects is not impossible, and such a world would have no events. Thus objects are metaphysically prior to events, for a condition of the latter existing is that the former exist. There are two responses to this objection. The first is to challenge the metaphysics. There is no reason, independent of a particular theory, to think that either the category of events or other categories (of objects, properties, or whatever) will turn out to be fundamental. Although I favour such a challenge, I will not defend it here, for it should be clear that a second response can be made in the light of what was said to the first set of difficulties. Since a unity theorist can make the distinction between an item and its life history without giving up the enterprise of explaining the unity of broad things, this second objection can be bypassed. An object's life history is a sequence of events;.the chain theory is concerned with the issue of which sequences of events constitute the histories of particularsunified, persisting things-and which do not. But, as we have already seen, a particular is not to be identified either with its spatial components or with its spatio-temporal spread. In effect, both of these objections have been answered by distinguishing constitution from identity-a distinction that is familiar to identity theorists. \3 As will become clear in the fifth 13

See e.g. Wiggins [7], Wiggins [9], and Hirsch [2], ch. 2.


Metaphysical Problems

chapter, constitution and identity will be distinguishable in ways that have not so far been recognized by most theorists. Let it be noted, however, that a difficulty confronts the theorist who would maintain that objects are distinct from their components, yet are in some fundamental way three-dimensional. MelIor, for example, has maintained that the three-dimensionality of things is so fundamental that, supposing the general theory of relativity entailed the denial of this, we would have to give up that theory. 14 A theorist like MelIor can give &.ome account of the spatial unity of things-either along the lines suggested above, or along quite different lines. Since any unified object is wholly present, on his view, at any given time, we can at least make sense of the question of what its unity consists In.

But if an object is wholly present at a given time, it is not clear what sense can be made of the unity questions I have been asking so far. How can there be a real question of unity over time for an item that is completely present (and thus present as a unified item) at a given time? Remember that the three-dimensional unity theorist gives an account of spatial-and only of spatial-unity. Here are all these regions of space occupied by particles, or other things. Some of these regions sum into regions occupied by larger, spatiallyunified things (desks, bridges, sandwiches, and so on). Even if the three-dimensionalist thinks that there is not interest or merit in theories of spatial unity, such theories make sense in that the puzzles. they answer can be expressed in tenns acceptable to the three-dimensionalist. But if we limit ourselves to the threedimensional conception of objects, we cannot even give voice to the unity puzzle. The parallel question asks us to think of all the timeslices of objects occupying space-time. Many of these slices sum to unified, four-dimensional things, and many do not. The chains which are constitutive of objects cannot even figure in the threedimensional account, for as long as we restrict ourselves to three dimensions, there just are no such chains. 15 14 Melior's protestations remind me of another Cambridge philosopher's claim that if the quantum theory delivered results contary to common sense then that theory would have to be abandoned. See Melior [1], p. 285. The issue of things and dimensions is explored in more detail in Ch. 5 of the present work. 15 It can be argued that three-dimensional objects can endure through time. Something like this seems to be Melior's position. But if we observe that what are usually called 'temporal' pans of objects are in fact spatio-temporal pans of them, we can begin to see the oddness in that position. If an item has only spatial pans, then the

Metaphysical Problems


This inability to pose the question typically forces those committed to the three-dimensional conception of objects into two further dogmas. One is the dogma of denying that certain puzzles actually arise. Wiggins, for example, considers that a certain kind of fission puzzle for persons might be tackled using a four-dimensional, mereological approach. However, rather than deal in mereological unintelligibilities, he is prompted to question whether the fission case described really is a conceptual possibility. 16 Here, I suggest, is a clear case of letting philosophical predilections carry too much weight. The fact that the fission puzzle can best be tackled mereologically will, after all, reassure the confirmed four-dimensionalist that the thrc;e-dimensional account of objects is inadequate. A second dogma of three-dimensionalists is the often repeated claim that identity is primitive and irreducible. It is thus incapable of explanation. I do not, as has been already explained, deny the primitiveness of identity. But what I recommend is that we explain the unity or identity of some things, and in explaining this we accept the unity or identity of certain other things. Relative to any theorizing we engage in, there will be a distinction between the primitive and the non-primitive. But there is nothing, as far as I can see, to show that identity is always primitive. It would be monstrous to maintain that the identity of cenain theoretical or social question of its extent in time does not arise, for time is not one of three spatial dimensions. Minkowski space-time is not simply Euclidean three-space with an extra dimension attached, as Lucas observes: 'Minkowski geometry is markedly different from Euclidean geometry-as we have noted, a line in Minkowski geometry can be at right angles with itself ... Moreover the (3 + 1) dimensional geometry of space-time is very different from a (2 + 2) dimensional geometry (two space-like and two time-like). It is therefore dangerously misleading to speak as though time were just a fourth dimension of space.' (Lucas [1], p.205.) cf. the disagreements in Nerlich [1] and Melior [1]. Nicholas Measor takes a similar line in unpublished work, where he suggests that an item cannot, strictly speaking, be in both spaces. However, my own line here is more cautious. What we have to recognize, as argued in Ch. 5, is that items may be viewed either three-dimensionally or four-dimensionally. But there are no good grounds for the claim that a particular thing is 'really' three- or four-dimensional. cf. also Lewis [6], ch. 4.2. 16 Perhaps I am too crude in attributing this move to Wiggins. Here is how he puts it: 'The conceptual possibility of a delta in the stream of consciousness jogs our whole focus on the concept of personhood. But, rather than jump to the conclusion that we have no idea at all of what we are about ... let us go back to the beginning and ask: Is such a delta really a conceptual possibility?' (Wiggins, [9], p. 169.) Measor is in no doubt, however, that Wiggins's strategy is as I -give it.


Metaphysical Problems

constructs-say ecosystems, income groups, social classes, or groups bound by certain kin relations-is primitive and incapable of definition. The objection to four-dimensional accounts of unity is thus not motivated by any general principle as to the universal primitiveness of identity or unity. Rather, it is a dogma applied to persisting things because of the fear that any explanation of their unity will involve four-dimensional notions. 17 None Of this, then, is a denial on my part of the primitiveness of identity. Relative to properties, taken as primitive, we can try to give an account of objects as bundles or clusters of properties. If we instead take objects as our primitives, then we can try to explain properties as sets of objects. There is no final way of settling the question: but which is the real primitive, objects or properties? Rather, we could try to adjudicate between object-based and property-based theories and ask: which is, for the time being, the better theory? Just so is it with identity. We will give the survival theory a run for its money in the next few chapters. But that means that stages will be our primitives, and so the identity of stages will be used in giving an account of the identity of broad objects. Of course, we could take broad objects as primitives and use them, as suggested already, in constructing a theory of stages. And then we could argue about which theory worked better, and which we should prefer. Either way, we are taking the identity of some things as primitive. Perhaps in this way, we can best appreciate the truth that identity itself is a primitive; though appreciation of this truth in no way diminishes the interest of constructing a theory of the unity, or identity, of persisting things.


Since so many theorists of identity speak, entirely naturally, of 'continuity', this term is bound to crop up later in my treatment. The problem, we now see, is how to understand it. For we have just discovered how conceptual development (as I will call it) can lead us 17 To be fair to some of my opponents here, it is wonh noting that some of them champion the primitiveness of identity on general grounds: they are what I later call Joundationalists about identity, holding that at least some identities must be primitive in order that non-primitive unities do have grounding. For a discussion of their position, see Ch. 8.9.

Metaphysical Problems


into a bind. If we develop the notion of continuity in a mathematically precise way, we get a concept that does not apply to many of the changes we normally count as continuous. However, if we develop the concept so as to embrace Hollywood-style 'continuity', then we can no longer distinguish the spatiotemporally continuous (whatever that means!) from the discontinuous. This finding is enormously gratifying to one who, like me, believes that we frequently operate with concepts that are extremely unclear, perhaps even incoherent (in the way Dennett suggests our concept of pain is) and yet tremendously useful/or all thatpe The term 'continuity' will be used in what follows to mark some intuitive notion of a continuous path in space-time which we cannot, at the moment, make clear in any mathematically precise way. This chapter has not, however, finished off continuity theory for good. In addition to the attacks on it made in this chapter, some time will be spent in Chapter 4 showing that a continuous path in spacetime (whatever that is) is not necessary for the unity of an object. Even such obviously persisting things as mountains can be unitary, while discontinuous. The considerations in that chapter really do put paid to the continuity account. However, it is worth avoiding a common pitfall in philosophical theorizing. We often try to establish the plausibility of one theory or account in a rather indirect way. First of all, the major rivals to the favoured account are shown to be false, implausible, incoherent, meaningless, or whatever. Having obtained by this strategy a relatively clear field for the favourite, no argument at all is given on its behalf. It is presented as self-commending given the failure of the rivals. In this work, I try to do rather more than this. As well as showing the implausibility of any continuity version of the chain theory, I also try to show that there are independent reasons for favouring a version of the chain theory using the account of survival already sketched. Of course, one thing we must remain clear about is the distinction between epistemological and metaphysical, or ontological, questions. What I mean is the distinction between our basis for judgements about things in the world on the one hand, and the way the world has to be if such judgements are true on the other. We IB See the article 'Why You Can't Make a Computer that Feels Pain' in Dennett [1].


Metaphysical Problems

must be very careful about how we put this matter. We might think we could phrase the identity or unity question like this:

Cl) I perceive a desk in my office at 3 p.m. (2) I perceive a desk in my office at 5 p.m. the same day. (3) Are the desks I perceive the same or not? I t is quite remarkable how many theorists who ought to know better fall into this kind of trap when putting forward their analyses of identity. Wiggins writes, for example: 'If one locates each of the particulars a or b (under covering concept or concepts) and ... traces a and b through space and time ... one must find that a and b coincide'19 as the truth condition for 'a=b'. But if a=b then there can be no question, strictly speaking, of locating 'each of the particulars a and b'. For there is only one persisting thing to be located, called 'a', and also called 'b'. Wiggins's mistake-an entirely natural one-is to talk in terms of two objects, instead of dealing with the case as one in which there is but a single object with two names. Our above case is similar. If I perceive a desk in my office at 3 p.m., then there is a desk in my office at that time and I perceive it. Likewise for the five o'clock case. But either there is one desk there all the time, in which case there is no question of desks being the same, or the desk I saw earlier has been replaced by another. In the second case, there is no question of the two desks being the same in the numerical sense, no matter how similar they may be. So how are we to phrase the matter? From the point of view of our experience, we can talk about experience-of-a-desk, where the experience in question is indifferent to the identity of the object in the world producing it. Suppose my experiences-of-a-desk in the office are both produced by a desk (though this is not to deny that I could have an experience-of-a-desk produced by something that is not a desk). Then we can ask: was the same desk responsible for producing both experiences, or were they produced by more than one desk? And, again, if we wished to complicate the issue even further, we could even ask if the experiences were produced by things other than desks. But let us keep to the simple cases for the moment. So we can now get clear on the epistemological issue. Our 19

Wiggins [2], p. 45.

Metaphysical Problems


experiences of the world are many, and usually short-lived. Of course, I have been working at my present desk for many years now: so I have had thousands of desk-experiences. My theory of the way the world is allocates to just one desk the role of having produced these many experiences. Incidentally, given university policies on furniture renewal, my theory may be quite false. For, on reflection, I realize that there have been opportunities over the years for my institution to have given me a different desk without my having noticed. Luckily, such occasions have been relatively rare: yet, of course, the slight probability remains that my judgement about the furniture is wrong, and not wrong for any philosophical reason. What, then, are the philosophical problems? Well, there are two. The first one can be put fairly vaguely. How is it, that on the basis of our hotchpotch of experiences we have constructed a picture of the world as consisting of relatively stable, persisting things whose changes are predictable and ordered? The answer to this great question is liable to have something in it about the nature of the world and the nature of we who experience it. Secondly, there is the question: what is it we are judging when we say that one persisting object has been esponsible for all my experiences-of-a-desk in my office, and a quite different object has been responsible for some other set of different experiences, and so on? Our business is to answer this second question by saying something about the unity of things; but our answer is bound to have some relevance to the first great question as well. Because empiricist philosophers these days often say strange things about our knowledge of the world, Chapter 7 is devoted to correcting some current myths about this. In that chapter, I argue that the world is structured in such a way as to make our knowledge of it possible. In terms of our second question, we have set out to discover if what we are saying about the world when we say it contains my desk is that it contains a sequence of desk-stages connected by the survival relation. It will not do to cast survival aside and opt instead for spatio-temporal continuity; for, in terms of problems, we have not so far encountered-nor will we encounter-difficulties with the survival relation to match those encountered in explaining continuity.

3 Structure and Survival 3.1. THE EUSA PROBLEM

Three points were sketched out in Chapter 1. First, we noted that survival can occur as a relation between distinct things. Second, it was suggested that the causal and structure conditions are each necessary for survival, while the matter condition added to the other two yields a trio of jointly sufficient conditions for survival. And thirdly, it was argued that for cases in which survival claims seem natural, our conditions are plausible. In this, and the following, chapter, we look at some distinctly non-normal cases. These are cases which do not arise; indeed, they may some of them be impossible to realize. We are thus engaging in an activity that is still fairly common in philosophy: the construction of thought-experiments. However, it is important to recognize that such experiments are not without risk. One rationale for using them is that we thereby employ concepts in ways that put demands on them far in excess of those made in daily life. Just as engineers subject mechanical components to strain tests that far exceed any strains likely to occur in their normal situations, a thought experiment supposedly enables us to expose concepts to loads not normally encountered. One problem, however, is that whereas the engineer can pronounce on tensile stength after the strain test, it is not clear what properties of concepts can be discovered from thought-experiments. The trouble is that in looking at significantly unusual cases we stray away from conceptual analysis and move towards what might be called conceptual development. For an every day example of what this phenomenon involves, think of how blind people describe their examinations of objects, or their meetings with people. Anyone who has a blind friend or relative will be aware of his or her use of 'see' in what strikes a sighted person initially as a quite inappropriate way. Yet, in time, it becomes quite natural to use the word in a way that is related to our ordinary, daily use of it, but significantly different, for it records investigatory successes without involving visual percep-

Structure and Survival


tion. This is a simple example of conceptual development, for our concept of seeing is not so well determined that we cannot develop it in this new way. Now, my suspicion is that analytical philosophers practise conceptual development more often than is usually admitted. Given the indeterminacy, and vagueness, in our concept of a person, fM example. what is meant to be an analytical study of personal identity may end up proposing one particular development of the concept over various rivals. The competing conceptions of personhood may fail to uncover one agreed core of doctrines for the very good reason that there may be no such core. Applying this thought to the case in hand, it is necessary to realize that I shall be arguing for particular conceptual developments in this and the following chapter. In Chapter 4, for example, I will argue for a specific development of the concept of physical object in the face of certain imagined oddities. So any confirmation for my proposed conditions that emerges from these experiments is itself conditional upon acceptance of the plausibility of the conceptual development in what follows. It can be helpful to think of conceprual development in terms of what we might call the Eusa problem. In Hoban's novel, the Eusa story is venerated by the primitive folk who are remote descendants of the survivors of a nuclear holocaust. The story itself is a strangely garbled mixture of terms drawn from the nuclear physics and politics of our day incorporated into a narrative expressing the predicament of these sorry survivors. The epigraph quoted on the flyleaf nicely represents the worrying circularity we face when dealing with either conceptual analysis or development. Eusa asks the little man (a mysterious figure in the story) how many changes there are, the response being that there are as many as required by the idea (we might say concept) of Eusa. Logically enough, the next problem is to find out what this cohcept or idea is-but this we will not know till we have seen all the changes. The exchange here reminds us of Aristotle's definition of substance as what admits of change. The point is that the changes appropriate to Aristotelian substances permit them to continue falling under the same concept. 1 We recognize that there are principles of change for many sorts of natural thing: indeed change is mandatory for trees I For Aristotle, 'sensible substance is changeable', while the substratum which persists though change is what he calls 'matter' (Metaphysics lk).


Structure and Survival

and mammals if they are to be properly counted as trees or mammals at all. The principles of development and growth for living things, incidentally, are the sort that would enable us to arrange a collection of Kripkean holograms into correct chronological order-for example, puppy-stage before adult dog-stage. But the changes that items undergo, or have undergone so far in our human experience, in this world, do not seem to exbaust the entire repertoire. It seems that we can readily enough imagine possible worlds in which things would still count as trees even if they behaved somewhat differently from trees as we know them. The concept of tree is indeterminate, or open-textured, enough, to allow it to apply in these other situations. We have seen just such a phenomenon in the case of the concept of seeing in relation to the blind. So this means that we could develop a concept from our presently vague concept of a tree which would apply to things (in some other world) whose behaviour is significantly different from the behaviour of trees as we know them. One thing the Eusa passage reminds us of is the difficulty of knowing where to stop. Philosophy must be different from science fiction. But it is hard to state just where in an exercise in conceptual development it is time to stop. A sensible methodological principle is that we do not continue conceptual development beyond the point at which it ceases to have value for our account of how things are in the real world. All philosophical theorizing-even that which involves quite different possible worlds-must keep its centre of interest in the one world in which we live and on which our curiosity is ultimately focused. A further insight we can derive from the Eusa problem is this. When tackling philosophical topic of a general kind, we tend to restrict our thinking to a particular range of cases. Then we simply project, or generalize, our findings from this range over cases we have not dwelt on. Alternatively, we just do not think about the other cases at all. So we must be wary of coming up with theories of identity or of survival that are fine for relatively unchanging things-like many artefacts, or mountains, or works of art-but which make the identity of much more changeable things mysterious. Locke was well aware of this difficulty, and insisted that identity was suited to the idea of the thing in question. 2 It follows 2 Locke [1], H. 27: 'It being one thing to be the same substance, another the same Man, and a third the same Person, if Person, Man and Substance, are three Names standing for three different Ideas; for such as is the Idea belonging to that name, such must be the Identity .. .'.

Structure and Survival


that an account of identity or survival for mountains may well not be suitable for persons, or vegetables. What is true of accounts of identity is also true of accounts of change. So Locke would agree with the little man that we need to know the changes an item is liable to go through before pronouncing on the principles of its identity. In terms of talk about kinds, or sorts, we can maintain that different possibilities of change are associated with different kinds of thing. The Eusa insight will be very much with us throughout this work. Even in this chapter, we will be speculating on the identity of daffodils-a topic strangely neglected byidentity theorists for, if I am right, attention to daffodils will help us in the end resolve some of the problems of personal identity!


The case to be studied in the present chapter involves a remarkable imaginary piece of technology which Dennett has called the teledone. In essence, teleclone machines permit transport of people and their belongings to distant corners of the universe without any movement of their physical stuff. How can this be? In its simplest form, the machine consists of two parts-a 'sender' unit and a 'receiver'. The sender unit dismantles the bodies of potential interstellar travellers swiftly and without pain into their molecular and submolecular components. During this process, however, the unit makes a blueprint detailing the structure and arrangement of every molecule and atom in the subject's body. When complete, this blueprint is beamed at the speed of light to the distant receiver unit where carbon based molecules are cleverly combined in accordance with it so that an exact replica of the traveller is constructed. Thus the teleclone differs quite significantly from the 'normal' transporter beam beloved of science-fiction authors. When thinking about the teleclone in this form, many people are inclined to accept that those who 'travel' in this way survive the experience. Of course, the machine can just as readily be regarded as what Dennett calls a 'murdering twinmaker'. 3 But it may be that talk of murder in this context draws attention to the lack of identity between the original traveller and the telecloned replica. It is easy to 3 See the introduction to Hofstadter and Dennen [1]. Parfit calls the same machine a 'teletransporter' (see Parfit [4], ch. 10).


Structure and Survival

see how the machine could be modified so that a sender unit transmits blueprints simultaneously to two distant receiver stations in quite different places. In this case, the process would lead to the production of two replicas-a kind of fission by teleclone. For obvious reasons (mentioned in our earlier discussion of Theseus' ship), we could not count the original traveller as identical with both replicas. But then, we might argue, identity is not present in the simple case either. This argument relies on a principle called the 'only x andy rule' and has been deployed to good effect by Bernard Williams. We will discuss it in the next chapter, when considering the role of rivals in establishing identity claims. For the moment, let us accept the Williams point. If we would not want to count the traveller identical with both replicas in our latest case, then let us take this as showing that the traveller is not to be counted as identical with the replica in the original case, noting that all that distinguishes the two cases is the presence of a rival for identity in the latter one. None of this, of course, counts against the survival claim. As we have already seen, interesting cases of survival occur in relating different things. In the first case of telecloning, the traveller may seem to survive as the replica; in the second case, the traveller may seem to survive twice over. Of course, we can hardly accept this reasoning just as it stands. For we might wonder if it is right to make even a survival claim in this case. Would we have any sympathy for someone who viewed telecloning as something to be avoided if at all possible? What the teleclone does, after all, is produce a replica of me. Such a person will be indistinguishable in all respects from me as I am at the time I enter the teleclone sender unit. But what comfort does the production of a replica afford me? It may be good news for those near and dear to me that someone indistinguishable in all physical and psychological respects will emerge at the other end (assuming, of course, that nothing goes wrong with the process). But should any of this afford me comfort? It seems quite plausible to maintain that our attitude to ourselves may be very different from our attitude to our pets, our favourite music, our friends, or even our loved ones. The thought that my replica will shortly be around on some distant planet, and that a replica of that replica will in five years be back here on earth may prove unconsoling to me no matter how high the quality of reproduction. Part of the trouble here is that we are handling a complicated case

Structure and Survival


before we have even become clear on the simple cases. It may be that in the end we have to admit to permanent puzzlement over the prospects for personal survival. This would not in any way inhibit us from examining the question of whether less complicated, lowerlevel, items might survive the teleclone. And notice, too, that remaining agnostic about the question of the survival of a person does not mean that we could never decide about whether to teleclone or not to teleclone. An agnostic might encounter a situation in which telecloning is undoubtedly the sensible choice. Indeed, even if I thought that death was a certainty for me in the sender unit I might very well choose to teleclone if I happen to be facing certain death in any case, stranded in some far-flung corner of the galaxy. I might consider the sending of the blueprint and the production of a replica a last act of kindness to those who care about me. What adds complexity to the situation is that we feel comfortable with the notion that high-level objects-like persons-have the sorts of properties that can survive destruction of their physical embodiments. No doubt, this kind of view is a legacy of our Christian-Platonic concept of the soul and its Cartesian development in the concept of mind. Nowadays, with analogies to computers in fashion, we are maybe tempted to think of persons as a kind of software: the package that constitutes the real me can perhaps be run on different hardware. What the blueprint carries in all its detail about my brain and central nervous system structures is a program which can be run on any suitable machine. Attractive though this way of thinking is, my strategy will be to suspend judgement on this matter for the moment and instead to concentrate on getting clear on the issue of survival for lower-level items. However, those who already find the idea of persons surviving the teleclone congenial should find my argument that lower-level items also survive it quite easy to accept.


Let us turn, therefore, to .the case of simple things, like stones, models, shoes, shrubs, and typewriters. Any of these that can be accommodated inside the sender unit can be telecloned and can survive as their replicas. Why should we say so? Every scratch on


Structure and Survival

the stone, every detail of the model, every ripple on the sole of the shoe will be reproduced by the te1ecloning process, and reproduced in exactly similar material, let us suppose, to the original. Just as in our case of the brother who copied his sister's model bridge, we seem to have clear grounds for making a survival claim. What more, we might wonder, could there be to the notion of survival? Of course, there are those who see no sense in talking about my survival unless by this we mean my survival as myself. Just as in the first chapter, I have to ask for such readers to treat my remarks on survival as stipulative, and hope that my analysis of identity by means of survival eventually convinces them that there is a use for the notion, however bizarre it seems to them. But a different objection may come from quite another source. For there may be those who are happy to concede that identity and survival are distinct. Further, they admit that it makes sense to talk of a person surviving the teleclone. But they deny that anything other than something that lives can survive the procedure. I and the rose-bush may survive; but my typewriter and my shoes do not. What motivates this new objection is a kind of Lockean distinction between living things and mere 'masses of matter'. For Locke, non-living things are 'mere cohesions of Particles of Matter any how united'. By contrast, living things are a 'disposition of particles' together with an organization of parts suitable for receiving and distributing nourishment and for continuing the life (vegetable or animal) of the thing in question. What te1edoning does is replicate this life-supporting organization, or structure; but such replication is what goes on in the normal course of events all the time inside living bodies. My pancreas renews its cells roughly every twenty-four hours; what keeps my pancreas the same throughout the changes is the replication each time of the previous cellular structure. 4 . Although Locke is one of the most interesting historical writers on identity, his distinction here seems to be misconceived. Suppose he happens to be wrong about the animate versus the non-animate. We can easily imagine discoveries in physics that tell us that crystals of quartz, despite their apparently fixed and stable structure, undergo frequent renewals of their subatomic parts. This new turn to science would effectively put living and non-living things on just • The reference is again to Locke [1]. n. 27.

Structure and Survival


the same footing as far as structure and organization are concerned. But we do not even have to go as far as this kind of speculation to establish Locke's errors. For he is wrong about how things are, given our current orthodoxy. Very few inanimate objects are, as a matter of fact, completely without internal structure and absolutely amorphous. It is the highly organized structure of materials like steel and iron that make them so suitable for industrial use and artefact manufacture. Different kinds of steel, for example, not only differ in respect of the amount of elements, like carbon, that they contain, but also in the alignment and arrangement of their atoms and molecules. Processes like annealing and tempering are specially designed to induce specific crystalline structures that give metals their useful and distinctive properties. But we can stop short of this fine level of description and still make the point about structural arrangements. The child's model bridge is hardly a cohesion of particles 'any how united', nor is even so simple a thing as a wooden carving. The pattern of the grain in the latter, like the arrangement of the blocks in the former, may be entirely distinctive. If the teleclone succeeds in reproducing such patterns and arrangements, then why should we deny that a model, or a carving, survives as its replica? What the Lockean objection has done, in fact, has been to bring to our attention the importance of our structure condition to survival. To the extent that structure is preserved in our replication process, then we have partial grounds for a survival claim. Of course, the teleclone satisfies our other conditions as well. We get replicas made in similar material to the originals, and we also have a straightforward causal account of why structure and matter are thus preserved.


It may have occurred to some readers that an item and its replica seem to be good candidates for being tokens of the same type. This is indeed the case, provided the standard of replication is high enough. Of course, many people have suggested that the type-token distinction can be made to apply to persons. As we will see later, Parfit's account of what he calls 'q-memory' gives us a good example of the applicability of the distinction to psychological items, and Bernard Williams has previously suggested a way of


Structure and Survival

understanding the idea of person-types. 5 Since at this stage of the investigation we are trying to avoid dealing in any detail with complicated cases, let us continue to restrict our attention for the time being to simple physical things. The type-token distinction is commonly used in linguistic contexts. For example, we distinguish utterance, word, or sentence tokens from utterance, word, or sentence types. In the last sentence, for example, the type word 'utterance' has two token occurrences. If we say that the one word 'utterance' has occurred twice, then our reference is to the type. If we say two different words have occurred in the sentence, even though they have just the same fonn, then our reference is to word tokens. Clear though the distinction appears to be, it is by no means easy to state it precisely, nor to give a formal specification of what a type is. One possible benefit of the account given here is that we can approach the definition of types and tokens by way of the theory of replicas. As I have argued elsewhere, artefacts are not only amenable to the type-token distinction, but in our conversations about them we consistently fudge just this very distinction. 6 This fudging on our part may partly explain why the distinction is so hard to state precisely. Such referential opportunism, as I can it, aids and abets the nonnal flow of conversation by letting us avoid over-specificity in discussions about cars, bicycles, cameras, or clothing. If I ten my friend that I am going to buy the same car again, then it is very likely that I would be construed as referring to a car type. This is not to deny that in some circumstances, we find ourselves buying the same artefact token more than once! We can think of my present car, then, as a token of a certain car type. In the case ofTheseus' ship, we can think of the continuous ship and the plank-hoarder's product as both of them being tokens of one and the same ship type. Applying this distinction to the case of Theseus' ship suggests that some of the problems raised by it may not be so acute as we See his essay 'Are Persons Bodies?' in Williams [5). For more details, see the original treattnent in Brennan [3). If what is argued later about the case of persons is correct, then we are similarly opportunistic in our discourse about them. Sometimes, in our valuations of persons, it appears that we are concerned with features of the token: I like someone's smile because it is the smile of rhar person. In other cases, I like someone's smile because I like smiles of those kinds or types. One difference between our valuings of persons and of cars is that it is easier to become aware of the existence of opportunistic referential practices in the latter case. 5


Structure and Survival


thought. Many artefacts are, as we might say, replaceable. If I break a part of my bicycle, or a piece of cooking equipment, then I will generally be satisfied if I obtain a replacement which is a fairly exact replica of the broken component. This is not, of course, always the case. Human beings can grow attached to, and fond of, even inanimate things. Thus nostalgia can make us reluctant to part with an old camera, or replace a tried tennis racket with a new model. In general, however, artefacts show a kind of replaceability which is not displayed by animals or other living things. (This last remark is controversial as far as some moral philosophers are concerned, namely those who would maintain that only persons are irreplaceable and that even animals are replaceable. Their views strike me as simple human chauvinism, but this is not the place to argue the point.) As far as the fuctioning of an artefact is concerned, the substitution of one artefact for another identical in structure and material should make no difference to the execution of the task in hand. Let us apply this thought, then, to the ship case. Suppose we have the two rivals before us: the Heraclitus and the Heraclides. Although they very likely differ in condition from each other, the ships may well look as if they have both been produced according to the same set of plans. And in a sense they have. When manufacturing large and expensive artefacts-say aircraft-it is not unusual for one, or several, prototypes to be built. These prototypes enable important aspects of the new design to be tested in real situations, away from the limitations of wind-tunnels and drawing-boards. If the design is a success, then the plans that enabled the prototype to be built will also enable further items exemplifying the same design to be assembled. But the prototype artefact itself can be regarded as a kind of codification of a certain set of plans. And we can regard the Heraclitus in very much this way. Its repair and the associated construction of the plank-boarder's ship is a kind of manufacturing process, albeit slow and inefficient. In a normal manufacturing case, we will know for sure which artefact is the prototype and which is the first of the production run. This is, of course, not the case with Theseus' ship, at least, not as long as we are unclear about the identity issue. But in the case of normal production processes, I can be interested in acquiring a certain sailing dinghy, let us say, while being indifferent as to whether I am buying the prototype or the first of the production


Structure and Survival

run. Indeed, if the prototype is in good condition after its trials, and is offered at a substantial discount, I may prefer it to the first of the production run. Likewise, if my interest in Theseus' ship is in getting a trip on a ship of its type, I may be indifferent as to choosing the Heraclitus or the Heraclides. If my interest is in purchasing one of them, and assuming that both offer much the same by way of seagoing qualities, accommodation, and so on, then I may choose the one which offers the best value for money. In each of these cases, my interest has been in the type, we might say, rather than in a specific token. Are there situations in which we might be interested in one specific token? It is possible to imagine that many circumstances require us to choose one token ship rather than the other, but even here there are surprises. Let us consider briefly the matter of insurance claims, or tussles over ownership. These legalistic situations seem to require a verdict on identity. Theorists who have given consideration to this son of case generally opt for a decision in favour of the continuously functioning ship: it is, and stays, the Heraclitus. There are good reasons for such a verdict. For example, it is easier to establish possession of a spatio-temporally continuous object (using the notion of continuity in the intuitive way); also, money invested in its maintenance and upkeep is lost if anything should happen to it (thus it is no compensation when the Heraclitus has been destroyed by fire if the insurers maintain that their cover in fact extended only to the Heraclides). These are, of course, simply pragmatic reasons for going along with the identity verdict. But it is wonh noting that to go along with the identity verdict in favour of the continuously functioning ship is not to have settled all questions of interest when tokens are under scrutiny. For example, suppose that in the refurbishment of the Heraclitus over time it turned out that cenain materials were replaced with different materials; some methods of construction were changed; new forms of jointing and caulking were tried; the rigging and sails were modified, and so on. All this is perfectly familiar in the case of artefacts like ships that require repair and refurbishment over a long working life. Now we imagine that long after both ships have sunk to the bottom of the Mediterranean, a naval archaeologist has the opportunity to raise one, or the other, but not both. Here is a case where a token has to be selected. But even if the verdict on identity has been given in favour of the continuous ship, our archaeologist

Structure and Survival


may find more of interest about the original methods of construction by lifting the Heraclides. When we think about this fiction, we can begin to see that we have a double success-as far as survival is concerned-in this case. This double success allows us very interesting choices-choices which can be independent of the decision, whatever it may be, on identity.


It is likely the case that whenever we have highly successful replication, we can talk of types versus tokens. Thus a good recording of a recording yields another token of that recording type: in this way pirate videos and other recordings are made. We have to take care, though, and not just about matters of copyright infringement. Suppose someone were to tape a record: the resulting tape might reproduce the performance on the original record, but it is not another token of the record, for tapes are not records. Records, viewed from the perspective of information theory, contain a recipe for the production of certain sounds given appropriate technology; this recipe is contained in the microgrooves cut into the plastic medium. Tapes can store equivalent information, this time in the form of patterns of magnetized particles. It is clear that the record-tape relation is much more complicated than other cases of replication we have so far considered. But one case we have looked at seems to cry out for the application of the type-token distinction. Surely, when we teIeclone a stone, or a rosebush, we produce thereby another token of the same type. Here is a result that gives us pause. For we do not normally think of the ordinary denizens of the physical world as types at all. Yet, if I had a teleclone machine, with sender and receiver units suitably close together, I could produce new tokens of any given object as readily as write new tokens of any given word. The introduction of the teleclone thus threatens to extend the duality of type and token to everything in the physical universe. Limitations on size would make it unlikely that telecloning could ever work for mountains and planets. But even they are not exempt from our reflections. So we have to contemplate the idea that a huge teleclone device could yield a replica of Mount Ev'erest from the original. Now imagine modifying the machine so that although the


Structure and Survival

object in the sender unit is scanned, and a blueprint made, it is not thereby destroyed (or perhaps it is disassembled and then quickly reassembled). Telecloning thus does not consist simply in the production of a sequence of tokens one after the other; it can allow the original to be retained, while more and more tokens are produced. Our ability to produce new objects of some same type is now much closer to our ability to produce further tokens of some type word. In the next section, we will consider an even more dramatic modification of the teleclone, but for the moment let us see if we can sort out the implications of this latest fantasy. We seem to have three things to deal with: replication, survival, and types and tokens. Let us see how far we can get using the notion of survival to yield an account of what it is for two items to be tokens of the same type. This will involve trying to use survival as an ingredient in the theory of replication. One consequence of this is that we may be able to render talk of types fairly harmless. This would be a move in the spirit ofPeirce and many who have followed him in declaring that types just do not exist. To talk of two tokens being of the same type, after all, is simply to indicate some relationship between them, rather than to make the ontologicallY bold claim that some further thing exists which they both instantiate. We will find that we can make a certain limited headway on the matter by trying to construe the relation between tokens as one of survival. We have already noted the importance of the structure condition in contexts of survival. It, together with the causal one, seems to suggest a mode of dealing with the type-token problem. Survival comes in degrees, and in some cases we can find a very high degree of survival between one item and another. Indeed, the brother's replica of his sister's bridge, or the new recording tape obtained from copying another precisely similar tape, are good examples of one item surviving as another to such a high degree that we are able to talk of replication. Just what do we mean here by talking in this way? A good example is found by studying natural replicators, like crystals and genes. In genetic replication, each replica of a gene is itself a gene, that is, it has the same kind of material in the same structural relations as the original; and it has been produced by a causal mechanism of which we have some dim understanding. Interestingly, genetic replication itself involves transfer of chromosomal material each time a new copy is made: this transfer helps to

Structure and Survival


keep the replicating process highly accurate. But we can imagine causal mechanisms for producing replicas, whether teleclones, or digital recording devices, which would also be highly accurate yet in which there would be no transfer of material of this sort. Let us consider, then, whether we might not give an account of replication purely in terms of survival. When one item survives to a suitably high degree as some other item then the latter is a replica of the former. Further, when one item is a replica of another, they are both describable as tokens of the same type. Will this do? Unfortunately, attractive though the suggetion seems at first, there are a number of counter-examples which prevent this tidy assimilation occurring. The existence of these counter-examples is highly instructive; for it reminds us that the fit between theory and reality is often a messy business. First, let us take replication. We could only maintain the account of replication in terms of survival to a high enough degree if we were to use the term 'replica' in a somewhat artificial way. For example, a model aircraft assembled from plastic components would normally be regarded as a replica of the real thing, provided it was an accurate scale model of some real aircraft. So there are cases of replication that do not live up to our requirements. Worse cases are in store, however, when we consider the issue of types and tokens.


Since our account of survival involves a causal condition, we apparently cannot use survival at all in giving a general account of types and tokens. For there is no reason why, for two items to be tokens of the same type, they should be causally related. Thus, I may doodle on an envelope here, while simultaneously, someone unknown to me on the far side of the planet produces another token design which coincidentally matches mine. Our two designs are thus tokens of the s~e type although neither was causally obtained from the other. It may be thought that the example is strange. After all, we do know of perfectly ordinary cases where we can produce marks on paper that match those produced by others unknown to us. What I type, for example, will contain word, sentence, and certainly letter tokens that match those produced by others busy at their machines. And even though my tokens are not causally


Structure and Survival

productive of yours, yet there is a good causal story that explains the agreement among typewriters and word-processors about what tokens of English letters should look like. We could say in such a case that all the tokens thus produced lie on the same causal tree. The same goes, of course, for handwritten tokens. But none of this will do to explain the doodle case. Doodling is not taught like handwriting, and it is hard to see why we would even start to look for any causal explanation of the adventitious replication were it not for a theoretical interest in using the notion of survival in the account of replication. Besides, suppose we find a lump of quanz on Neptune that exactly replicates one here on earth. The behaviour of silicon and oxygen atoms no doubt explains in both cases why the lumps have the crystal structures they do: but we are not forced to trace both lumps back to some common ancestor, or otherwise locate them on some same causal tree. Although I have already used the term, reflection may make us doubt whether such adventitious matching should be described as 'replication' at all. In fact, it is not clear what to say about usage here, and the attempt to impose theory on the matter may itself arouse legitimate suspicions. Here are some suggestions. One reason, I conjecture, why we are suspicious of calling our bits of quartz replicas of each other is that we tend to take a replica as something produced by a copying process (about which more in a moment). Now, in the case of the matching doodles, we can make sense of these being said to replicate each other because we can envisage circumstances such that the one could have been copied from the other. Again, printing can be copied, as can handwriting. So when we look at two printed documents which contain some of the same letter tokens we can again speak of replication to the extent that one set of tokens could have been copied from the other set of tokens. Think now of type and token cars. Since production cars are reproducible, they can also be copied (although, as we will see in a moment, they are normally produced by a process that is quite different from a copying process). Only the introduction of the teleclone, however, has made mountains, trees, lumps of rock, and the like reproducible. We don't normally think of two voles as tokens of the same type or as replicas of each other-no matter how similar they seem to be. Barring processes like cloning, mammals are just not reproducible: two very similar voles are just two very

Structure and Survival


similar members of the same species. We are only able to think of the Neptunian quartz example in tenns of types and token because, I suspect, we had already prepared the ground for this by speculating about telecloning. But without that background we would not have thought of the lumps as anything other than two very similar (in ordinary talk, 'identical') lumps of quartz. If all this is accepted, then it seems to make sense to apply our account of survival to the type-token issue. If we go along this route, we part company with much contemporary wisdom on the issue, according to which type-token talk is simply understood to apply to similarity of pattern. 7 It is a problem, of course, for this current wisdom, to explain why your last inscription of an 'e' and mine, although not very good replicas of each other, still count as tokens of the same type, while 'identical' badgers or 'identical' panes of glass do not. Perhaps, then, we can agree that there is more to two items being .tokens of the same type than simply similarity of structure. The problem remains over how to specify the missing ingredient in the story. We could start by trying the following three conditions as sufficient for x to be a token of the same type asy (that is for them to be replicas of each other). First, x must possess a structure that is highly similar to y's. Second, x is materially similar to y. And, thirdly, it is possible to obtain one of them using a process in which the other has a suitable causal role to play. Intuitively, the most likely such causal role is that one acts as the prototype, model, or original from which the other is copied. These conditions obviously cover the case where one is in fact causally obtained from the other. In such situations, where one thing is in fact produced using the other, we have survival in our sense. So the proposed modification to deal with adventitious replication looks quite simple. We are saying two items are tokens of the same type where one survives to a suitably high degree as the other or where one could have been the survivor of the other (to a suitably high degree). We can now close this section by looking at some counterexamples to the theory of types just given. The first deals with personal survival, and we can look briefly at this even though we are not yet officially in a position to deal with such complicated cases. 7 There are a number of Analysis papers on this topic, including Simons [1] and Goldstein [I].


Structure and Survival

On a remote planet, in a distant star system, there is a person who is an exact replica of someone-let us call her Beatrice-who lives here on earth. Indeed, if telecloning existed (which in this story it doesn't) Beatrice's twin is exactly just what the teleclone might have produced, as far as superficial appearance is concerned. Beatrice dies tragically young. But her replica lives on. According to the theory just given, we seem to be able to make sense of Beatrice having a replica, but apparently have to deny that Beatrice survives as that replica. For although Beatrice's twin could have been produced by teleclone, she wasn't. But should we not perhaps think again? Is it not plausible to suggest that in this case Beatrice does, after all, survive? The objection, therefore, suggests that we need to modify the account of survival. Instead of saying that in some cases of replication, one item could have been the survivor of the other, we now say that all cases of replication are cases of survival. And our causal condition on survival now requires a modal clause in it: if x survives as y then it is possible to obtain y using an appropriate process in which x has a causal role to play. Now this objection may have the thought underlying it that if, per impossible, Beatrice's twin could be transported to earth, those who had known Beatrice might be fooled. And, once they knew the circumstances, they might still want to regard Beatice as surviving as her twin. There is, no doubt, some force in this worry. But the objection is not conclusive. IfBeatrice's twin had been produced by the teleclone, then Beatrice would have survived as her twin. But she was not so produced. We seem faced, then, with a choice between maintaining a strict, non-modal account of survival, which wiIl make that relation incompatible, on occasion, with replication, or weakening our account of survival, so that survival and replication simply coincide. A decision either way is likely, in the end, to be stipulative. For, interestingly, we are faced, even within our theory, with the possibility of development, rather than analysis, of the concept of survival. 8 But we need not turn to science fiction to find cases where replication is incompatible with survival. For we need to notice that 8 This development would, of course, threaten our strong causal ingredient in survival; since survival can be regarded as constitutive of identity in some cases, then the proposal also threatens the claim that unity is a causal notion. For some people, this would be good news. See the discussion of Schlesinger's objection in Ch. 8, n.29.

Structure and Survival


there are in fact two distinct ways in which causal processes may be involved in the obtaining of replicas. We can dub these copying processes and production processes respectively. And only in copying processes do survival claims seem appropriate. In a copying process what happens is that a certain item is a model, original, or prototype from which a replica is made. By contrast, in a production process, a number of replicas are produced simply by going through the same causal process (often on similar material) a number of times. The last six stitches made by the sewing machine, for example, will be replicated by the next six so long as we do not change the thread or adjust the stitch. length or needle swing. The earlier stitches, however, do not participate directly in the production of later ones. They thus do not survive as the later stitches. Again, notes produced on a piano, or jam jars in the factory, replicate each other provided the producing causes are not varied too much. In this way we can get a whole range of cases of replication that are rather different from the photocopying, tape-recording, and telecloning cases we have studied so far. And, like our cases of adventitious matching, the existence of such cases threatens the role of survival in the general account of types and tokens. The safe conclusion thus seems to be that we can still avoid the Platonism that was threatening our earlier account of objects as types. But the story of how to avoid this Platonism is getting increasingly complicated. In the case of copying processes, to say that two things are tokens of one type is to say that one survives to a suitably high degree as the other. In the other cases-those where production processes are involved-items that are tokens of the same type are items of similar structure and (usually) matter produced by some same causal process. In this second case, we can have replication without survivaU None of this is to say that the fact that two items match each other 9 To try to deal with the modal cases of replication, we might wonder if rwo adventitiously produced tokens are of the same type when either (1) one could have been the survivor of the other by virtue of some copying process, or (2) one could have been produced by the same causal process that produced the other. It is not clear that even these conditions will do for handling the Beatrice case. But they do offer the prospect of handling many of the cases of type and token talk that are common in the literature. To forestall a misunderstanding, it is worth observing that production processes may be involved in copying processes themselves. Moreover, a real production process (e.g. in a factory) may well involve both copying and production processes.


Structure and Survival

may not be of interest even when one does not survive as the other. We have seen with the Beatrice case, that a replica of Beatrice might be almost as good as a survivor, even when Beatrice does not survive as her replica. Likewise, I may want to choose a replica that is as similar as possible to the broken statue, even where the broken one plainly fails to survive in, or as, the replica. If both statues are reproductions of some original, then the rationale for such a choice is clear: for the original will have survived as both (at least in ideal cases). The rationale for the decision in Beatric;e's case is perhaps less clear. But we will return to this problem when we come to the study of personal identity. For just now we will stick with the austere conception of survival. However, note the cost that our austerity incurs. Instead of introducing strange entities, we have had resort to talk about possibilities· and possible worlds. To some empiricists, then, our victory over Platonism hardly seems significant. For them, we have jumped out of the frying-pan of Platonism into the fire of Aristotelian essentialism. lo For those, however, who see resort to possible worlds as harmless-at least when it is undertaken in a suitably modest spirit---our proP>q$als may have some merit. The conclusion of this discussion is that we must withdraw our support for the use of the theory of survival in giving a general account of types and tokens, and of replication. What looked at first a good idea has turned out to be fraught with problems. However, obtaining a decent, austere account of type-token language would have been, at best, an incidental benefit of the theory of survival. More significant for the project in hand is the discovery that the use of type-token language in some instances of replication is easily glossed using our account of survival: the tokens resulting from telecloning and other high-quality copying processes will certainly be new items of the same type as their originals.


Pseudomorphism is a quite astonishing natural phenomenon. As is well known, most materials occurring naturally, and nearly all the 10 In the light of what is said about real possibilities (Ch. 5), this worry might be thought to be somewhat overscrupulous.


Structure and Survival


materials we use in industry (with glass as a notable exception), have a well-organized crystal structure. A crystal is built up by repetition of certain basic structures-the so called unit cell of the crystal. Of course, these basic structures come in a fair variety of arrangements. The patterns displayed by solids on the surface of the earth are the results of the cooling of the earth from an earlier, liquid form. And in this slow cooling, atoms of the various elements had time to arrange themselves into a finite number of stable patterns. Under high pressures, and over time, changes in the distribution of atoms and electrons in a solid can still occur. Indeed, the ability of materials to change their crystal structure when far from their molten state is exploited in common industrial processes. Annealing, for example, consists in heating a metal to a temperature between a third and one half of its melting point. At this temperature, significant migrations of atoms can occur and new crystal structures form. II Pseudomorphism is one such change in nature, whereby, for example, the atomic ingredients of fluorite (calcium and fluorine) are replaced gradually by atoms of oxygen and silicon (the ingredients of quartz). Although the unit cells of quartz and fluorite are slightly different, the overall shapes of the crystals of both materials are much the same. Thus pseudomorphism is an example of preserving overall structure while seemingly changing constituent matter. With such a natural phenomenon as a model, it is not hard to dream up a modified teleclone machine that does the same kind of thing for lots of materials. After all, we can maintain the large structures of an item while producing a copy of it in material which has a very different fine crystal structure. So, if nature can change fluorite to quartz, while maintaining overall shape and dimensions, can we not imagine a teleclone that transmutes an object of one material into a replica of quite different material? Instead of using the teleclone as dubious transport over interstellar distances, we can use it for transmuting things here on earth into replicas in new material. Do you fancy transforming your rusty old car into one made of pure titanium (recognizing that, of course, it will no longer run)? Or how about seeing that bronze statue in gleaming silver? Even natural objects, like pieces of rock, can undergo wonderful


See Ian Wilson [1], 7.7. Pseudomorphism is described in Ames [1].


Structure and Survival

changes in their material stuff. The modified teleclone is no less than the Philosopher's Stone! The question we have now to face is whether particular things could be said, properly, to survive this new telecloning procedure, 'telemutation', to give it a name. Presumably, the new machine works on similar principles to the old ones. Let us suppose that objects placed inside it are destroyed, as in the original model described by Dennett. Computer crystallography, and goodness knows what all else, enables us to produce at the receiver terminal replicas with new atomic components arranged in appropriate crystal structures. So long as we have a large enough supply of atoms at the receiver terminal, and avoid nasty chemical incompatibilities, we can carry out the sort of transformations described. And surely we have not lost the old objects completely. They may have been destroyed. But they survive as, or at least in, their replicas, and-if we can agree on this-it confirms a significant discovery already suggested in the first chapter. For it is clear that structure-is more important than matter when it comes to survival. This conclusion will be reinforced by our Chapter 6 discussion of structure and stuff. We have perhaps now reached the limits of conceptual development in our study of the teleclone. It has prompted us to reach certain conclusions-conclusions that we might have reached by more sober, less fantastical, thinking. However, it has been more fun getting to these conclusions by the present route. Thinking about the teleclone has enabled us to get clearer on the type-token issue. Using the vocabulary of types and tokens in the case of Theseus' ship has helped us see that identity is not-for many purposes-what matters. What matters, rather, is survival. And thinking about the varieties of telecloning has enabled us to reinforce our earlier proposal that preservation of structure is of fundamental importance to survival. Survival, of course, is still itself something of a vague notion. It comes in degrees and although we might have thought we could give an account qf it in terms of matter, structure, and cause our latest thought-experiment has cast some doubt on the necessity of the matter condition. Preservation of matter may only be necessary for survival that is higher than a given degree. Those who are suspicious of thought-experiments and science fiction in philosophy might care to reflect on how we could have

Structure and Survival


reached these conclusions without them. In particular, it is worth noting that our latest use of the teleclone fantasy was simply a device for eliciting verdicts on cases that are parallel to a natural one. Here is a crystal offluorite, with a determinate shape, a certain number of facets, in contact with minerals which, over the years, will exert.the right effects to induce pseudomorphism. In time, our crystal of fluorite will be destroyed, as atoms migrate to be replaced by quite new ones. But, after its demise, our crystal survives in the new crystal, this time of quartz, which has taken its place and preserved its overall structure. Our teleclone fantasy simply allows us to corroborate this verdict. In this case, as elsewhere, our fantasies do not interfere with our sense of the real. Rather, they act merely as heuristic flavourings, adding spice to our analytical diet.


Identity and Discontinuity 4.1. PREVIEW

In the present chapter, there is an attempt to achieve two goals that are of significance for the work as whole. First, I try to discredit the claims of the continuity version of the chain theory, thus leaving the survival story with a clear field. Second, I argue that identity or unity is a causal notion. These results encourage the hope that we can give some account of unity in terms of a chain of stages linked by the survival relation. Although such an account may appear to be reductive, I will try to show that the account does not force us to give up deep intuitions about the primitiveness of some unity concepts-in particular certain notions about the concept of a person. However, dealing with the issue of reductionism is not easy. As suggested earlier, there is more than one way of thinking about particulars. The recognition that this is so is itself a powerful consideration against reductionism. Any account we give of, let us say, the three-dimensional unity of a particular in no way reduces all talk about particulars to discourse in purely spatial terms. At the start of the present chapter, a last attempt is made to convince those wedded to three-dimensional conceptions of things that there are perfectly proper four- and even five-dimensional ways of thinking about objects. A useful corollary of this result is that for theoretical purposes we may draw upon whichever conception of particular is most useful at a given time. The chapter continues by delivering a powerful attack on continuity theories, an attack based on the argument that particulars might be discontinuous in space and time without this preventing our assigning stages to larger, unified wholes. What supports the assignments of stages to a persisting particular in a world of discontinuities is precisely the causal nature of the unity relation. Some objections to the proposed case are considered, the most worrying of which involve the notion that causal mechanisms could not operate in the case of discontinuous objects. This objection is

Identity and Discontinuity


not disastrous, and the Humean account of causality is argued to have application to these unusual cases. Next, there is a brief anticipation of the later application of the results on episodic objects to mental states. For those not attracted by the claim that one enduring particular may have a discontinuous existence, it is shown that the version in terms of survival supports an alternative account of the discontinuity cases. This account seems more cumbersome than my preferred one, but is possible. On either account, it is clear that the theory of survival has an important role to play. The most beguiling objections to best-candidate theories of identity are then studied, and it is shown that these do not force the abandonment of the claim that rivals can make a relevant difference to identity. However, important relations-like the survival relation-are to a certain extent independent of identity. If we make a clear distinction between constitution and identity, we can avoid the identification of particulars with their histories or their paths through space and time. Finally, the chapter concludes by following up the Eusa insight. Different possibilities of change are associated with different kinds of thing. Additionally, it is shown that different parts and components are of varying importance to the determination of survival.


We started with a straightforward commitment to a four-dimensional view of the world, at least as far as our theories were to be concerned. The chain theory took objects as collections of temporal parts, themselves items with four dimensions-three in space and one temporal. Yet in the last chapter we used examples of survival that were not in keeping with the chain theory at all. We took the teleclone to be something that worked on one broad object-a person, table, or car-to produce another broad object. So the survival claim we made was not that a stage of a table survived the teleclone as a stage of (some other) table. Rather, we maintained that a table undergoing telecloning survived as another table. We have already distinguished holders of the chain theory from


Identity and Discontinuity

those who maintain that a complete particular thing (chair, person, or vole) is present at any given time. The three-dimensionalist, in Wiggins's words, never uses terms like 'boy' so that they' ... denote either "phases" of entities or (if that were different) the entities themselves frozen at an instant. They denote the changeable changing continuants themselves, the things that are in these J5hases." A natural development of this view, as we have seen, is to hold that in some way a whole object can be before me at a time. Thus, on such a view, all of my desk is in the room just now as I write. It is not clear that Wiggins wants to go as far as this, but some three-dimensionalists certainly do. 2 For such a thinker, not all of my desk would be in the room just now if, a few moments ago, I had carried one of the drawers into the room next door. By contrast, the chain theorist insists that, even if all the spatial parts of my desk are in the room just now, only a small temporal part of it is. The three-dimensionalists have apparently got common sense on their side. Yet the chain theory has numerous theoretical advantages. For example, it is very easy to put our unity question in an unconfusing way as: under what circumstances do a number of different stages belong to the history of one object? There is no comparably easy way for the three-dimensionalist to put the unity question. At first, it might look as if we could put it as follows. Given that one object (my desk, say) is wholly in the room just now, and given that one object (my desk) was wholly in the room last week, what circumstances ensure that it was the same object on both occasions? But this is no way to put the question that puzzles us. For the very form of the question suggests that it is my desk which was in the room on both occasions. If, as an alternative strategy, we try to put the question in terms of the relation between my-desk-Iastweek an'd my-desk-this-week, we get no further. For, although there are two ways of understanding this formulation, neither is very helpful to the three-dimensionalist. First, we could take 'my-desk-Iast-week' as designating a temporal slice of my desk. But this precisely what the threedimensionalist finds a puzzling notion. Moreover, if we take it that there are such things as slices, or stages, of objects, then last week's Wiggins [9], p. 26. See MelIor [2], ch. 8. MelIor does, in fact, allow that there are genuinely four-dimensional things, e.g. events. But he balks at counting ordinary physical paniculars as being genuinely four-dimensional. I


Identity and Discontinuity


slice of my desk is not the same thing as this week's slice of it. A great deal has been written about identity through time, or 'diachronic identity', in apparent indifference to this simple point. Of course, we know what the chain theory can say about the relation. between these different desk slices-but to say this is to resort again to our familiar four-dimensionalist strand of theory. Alternatively, we could take the phrase 'my-desk-Iast-week' as designating nothing other than my desk as it was last week. This successfully avoids problems about stages, but only at the cost of not allowing us to say anything about the identity of the desk except in terms of the identity relation itself. For my desk as it was last week can be none other than my desk as it is now (or any other time), and the relation here is simply identity. Yet it is precisely this relation-identity, sameness, unity-which we are trying to explicate. In an ambitious series of unpublished arguments, Nicholas Measor has gone rather further in making life difficult for the three-dimensionalist. 3 He has tried to show that identity over time-diachronic identity-poses a significant problem for the three-dimensionalist. For if there are relations that are criterial of such identity over time, he argues, they are bound to lack properties that the three-dimensionalist assigns to relations that could be criterial for classical identity (properties like transitivity, for example). Thus, in the literature on personal identity, Parfit has made a great deal of relations like psychological connectedness and continuity (relations we will be discussing in detail later). Measor's argument is that since such relations lack transitivity, they cannot constitute criteria of identity on any stringent, three-dimensional view of persons. I am not sure that pressing these kinds of point would really impress the three-dimensionalist. Nor does it help to dwell on technical points-for example, the fact that Minkowski space-time is not simply a Newtonian three-space enriched by the addition of a further Euclidean dimension. The impact of this particular finding is, of course, dramatic. It means that items in Newtonian absolute space cannot be truly said to be in Minkowskian space-time at all. For those liable to be influenced by such considerations, the ., The arguments here are additional to Measor's arguments in his 'Four Dimensional Man' (see Ch. 2, n. 15).


Identity and Discontinuity

discussion and references in note 15 to Chapter 2 contain enough material, I hope, to help them make up their minds. I have to confess to a certain reluctance to make my philosophical theorizing depend too strongly on accepting any specific scientific view. This is not because I wish to deny that philosophy involves a posteriori theorizing-for it undoubtedly does. Rather, I want to continue to talk of space and time in a relatively informal way, while ensuring that what I put forward is not incompatible with our best attempts to understand the world in terms of current physical theory. As suggested in Chapter 2.5, the issue is not whether objects, continuants, persistents, particulars, things are really three-dimen.sional or not. On my account, there is a certain ambiguity in each of these terms. If we want to know how much space an object occupies, then we can consider it as three-dimensional. If we want to know how much spa~e-time it occupies, then we regard it in fourdimensional terms. And these considerations do not exhaust all we may be interested in knowing about an object. A balloon, for example, might have been blown up larger, or burst sooner, than in fact it did. We can contrast its actual three-and four-dimensional extent with other, possible, extents it might have had. A parent may regret the wasted potential of the child who might have been a brain surgeon: the possible careers of their offspring often loom just as large in the minds of parents as the actual careers. This issue of alternative possible histories will be discussed further in the next chapter. By then, it should be clear why unified objects, or particulars, in some ways transcend their parts, their dimensions, and their life histories. This very transcendence makes them rather elusive things. At first, nothing seems clearer than the idea of a unified, concrete, specific thing-like the lamp sitting on my desk. Yet as we begin to recognize that it has three-dimensional extent, four-dimensional extent, and possibilities that may never be realized, we can start to become worried about what the lamp itself really is. Is it something abstract, rather than concrete? Is it something that is the bearer of properties, while distinct from the properties it bears? We can avoid the resort to bare particulars and other metaphysical snares fairly readily. It is important to recognize that unitary objects-particulars-need not be the source of any special puzzlement if we are clear on the dimension-relativity of what is true of them. Let us go back to Gibbard's example of the statue and

Identity and Discontinuity


the lump of clay (mentioned in Chapter 2.5). Although these are distinct things, they can both be described as constituted of clay bits. How, then, can we describe the difference between the statue and the lump of clay? One obvious difference involves structure. To be a statue is to be structured in a way that a lump of clay need not be; for the same lump of clay can take up many fonns, while a statue is destroyed if its structure is significantly changed. For a lump of clay to constitute a statue, then, it has to have components that are organized in relation to each other. Indeed, one mistake Gibbard makes is to identify the statue and the lump of clay, while it seems clear that although the lump constitutes the stuff of the statue it is by no means one and the same thing as the statue. 4 Taken threedimensionally, then, there is no more to the statue than its components and their structure. To say this, of course, is by no means to deny that the statue is distinct from the aggregate of clay pieces constituting it. The lump of clay, however, taken threedimensionally, is no more than the aggregate of clay pieces constituting it (provided that by 'aggregate' we mean 'collection of suitably joined clay pieces'). Nor is the life history of a lump of clay anything other than the life history of an aggregate of clay pieces. Taken four-dimensionally, the statue is more than just a structured assemblage of components (themselves structured aggregates of clay pieces). Over time, the statue may lose a hand, or have a damaged leg replaced, while remaining the same statue (I simplify here: more doubt and less dogmatism will attend the discussion of artefact identity in Chapter 6.4). If the view urged later in this chapter is accepted, the statue over time will be a chain of stages linked according to the conditions on survival. But, taken as an item that might have had different histories from the one it in fact had, the statue is still more than a chain of structured statue slices. For this reason, we have to distinguish the statue as it actually is (which is a chain of slices, each of which is a structured 4 Resistance to my claim here is liable to be based on too 'concrete' a conception of objects or paniculars. Notice how misleading it is to describe a figurative statue as simply 'a lump of clay'. Of course the statue is, constitutively, made of that very clay and no other; but this is by no means to establish that it is a 'mere' lump of clay. Misleading, or false, descriptions, incidentally, are not always infelicitous. If the intention is to denigrate, then it is perfectly proper (pragmatically) to refer to a statue as a lump of clay, or to a car as a heap of rust.


Identity and Discontinuity

assemblage of components) from the statue as it might have been. So although the statue actually is a chain of slices, it might have been a chain of rather different slices. Viewed from the perspective of alternative possible histories, the humble lump of clay is more than a mere aggregate of clay bits. For the lump might have endured for longer or less long than it actually did. Even lumps have unrealized potential. So although the actual lump of day can be identified with a certain chain of lump slices (themselves simply aggregates of the same clay pieces) the lump of clay viewed as something with a number of possible histories is distinct from the chain that constitutes its actual history. Thus, as suggested previously, I can make the distinction between an item and its dimensions, its extent, its components, its trajectory or its life history, though for a rather different reason from that usually advanced. In an old piece on spatializing time, J. J. c. Smart comes close to representing what I think is the sensible view here: 'What we can achieve with a four dimensional logic, ordinary language achieves by using words like "cricket ball" with neither a three dimensional logic nor a four dimensional logic, but with a hybrid between the two. That is, it uses "cricket ball" with a three dimensional logic modified by the use of concepts like "change", "alteration", "staying the same".'5 Smart stops short of thinking about a five-dimensional account--one which we will explore in the next chapter. It should now be clear that a purely three-dimensional conception of particulars can only be maintained at some considerable cost. One of these costs might be a denial that the project followed in this book is possible at all. I have so far resisted saying much about criteria of identity, though they will be the subject of some later discussion. By the end of the present chapter, however, it will be clear that the search for conditions of identity is not only possible, but profitable and illuminating. That in itself will show a considerable advantage in the four-dimensional treatment of particulars. As a kind of reconciliationist move, I would point out that I have so far been happy to speak with the three-dimensionalist, even though theorizing with the chain theorist. This is a simple consequence of the fact that nothing is wrong with either the three5

Smart [1], p. 165.

Identity and Discontinuity


or four-dimensional treatment of particulars. Indeed, it is important to the methodology of the treatment here that we can take the survival claims of the last chapter at face value. Our success in talking about the survival of particular things (from the threedimensional point of view) can be taken as a basis for projecting this talk over the four-dimensional treatments. Our theory of identity and survival has to coexist, in the end, with the common sense way of putting things, even if-as happens elsewhere in the sciencesour official view involves some correcting of our ordinary way of expressing ourselves. In this way, conceptual development can lead to conceptual correction as well. Later in this chapter, we will see that the problems posed by the treatment of discontinuous things might lead us to reinterpret our everyday ways of speaking about identity. The present chapter, then, extends the theory of survival by showing the role it has to play in the unity of particular objects. We also choose a case that is intolerably difficult for the continuity theorist to handle, thus enabling us to focus on the importance of survival without the distraction of continuity considerations. As a result, the case for the view that survival is what is important to unity will be strengthened.


'A single object', wrote Hwne, 'plac'd before us and survey'd for any time without our discovering in it any interruption or variation, is able to give us a notion of identity' [Treatise, I. iv. 2]. Such a claim might fit well with the notion that the basic furniture of the world consists of objects displaying spatio-temporal and qualitative continuity. It is hardly surprising, then, that Hirsch should say: 'there is the most intimate connection between our concept of the identity of a specified sort of body and the idea of a spatiotemporally and qualitatively continuous succession of body-stages of that sort.'6 But what if-astonishingly-a new finding were to suggest the existence of a gap between our concept of identity and the way objects 'really' are, a gap associated, ironically enough, with a certain gappiness in objects themselves? • Hirsch [2], p. 235.


Identity and Discontinuity

Among macroscopic phenomena, there seem to be few examples of concrete, but discontinuous, objects (using 'concrete' to apply to items that occupy determinate volumes of space). Admittedly, pains and certain other 'inner' states, letters (that is, epistles), plays, television serials, and the like apparently exhibit identity along with spatio-temporal discontinuity. Yet none of these items is concrete, and the periodic nature of such astronomical curiosities as pulsars, variable stars, or transients is thought to result from the vagaries of observation. Thus, for example, a star in a binary system may appear to wink off at regular intervals, but only because the earth lies on the plane of the orbit of the two bodies, so that the invisible partner eclipses the other from our point of view. Perhaps, though, we are looking in the wrong place. Maybe pain tokens and tokens of mental states in general are concrete, while the types that correspond to them are not. This view would be taken by those who regard the concrete as that which can enter into causal relations with other things (recall our previous distinction between senses of 'concrete'). Numbers, on this conception, are nonconcrete, while thought-tokens are concrete. In the end, we may want to agree that mental events of many sorts are episodic, and concrete in just such a sense. But it is safer, for the nonce, to work with items of a somewhat low-level kind, namely those things that are uncontroversially space-occupiers, and ignore the duality of hardware versus software. It might seem that such things as orchestras and committees have determinate physical location and the sort of gappiness displayed by radio serials. But we must not confuse the episodic with the functionally discontinuous. Whatever reservations attach to the label 'functionally discontinuous', it is meant to pick out a class of things which admit disassembly while the components show continuity in space and time. Garden sheds, gliders, and even steamers, as we have seen, can be discontinuous in this way. When an orchestra disperses after rehearsal, or a committee after a meeting, we have a process similar to artefact disassembly. But committee members and orchestral players presumably show as much (or as little) continuity in space-time as the components of these other artefacts. So, for radical discontinuity, let us resort to fantasy. A series of studies involving high-speed photography and so on has been carried out, let us suppose, in order to check for brief discontinuities

Identity and Discontinuity


in ordinary objects. Amazingly, one of the hills adjacent to the Stirling campus has been established to 'flicker' occasionally. The hill (called 'Dumyat' incidentally) runs steadily, so to speak, for 167 hours, winks off for approximately l~ of a second, runs for a further 36 hours, winks off briefly again, returns to the 167-hour cycle, winks off again, and so on. Many simple tests, let us further suppose, have established that the hill is not simply invisible du;ing these brief flickers. Projectiles, for example, can pass through the place normally occupied by the hill provided they traverse it in the requisite l~ of a second; otherwise, of course, they become lodged deep in its interior. So there are times when the hill, however briefly, is just not there at all. Walkers, and the rest of us, however, are quite unaware of these changes, even though it means that on some occasions I may have been standing unsupported for those brief instants many hundreds of metres above the nearest solid ground! We may suppose that only a few such gappy objects have been found. The discovery suggests, though, that for all we know every physical thing is subject to regular even if brief discontinuities. Unfortunately, the continuity theory of identity seems no match for such a possibility. Whatever else is true of the Dumyat case, our experiments have revealed the existence of a number of objects successively inhabiting the location of Dumyat. Diagrammatically, the situation is like that seen in Fig. 9. Our problem is to relate our beliefs about


L -_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

Fig. 9.



Identity and Discontinuity

Dumyat to the succession Dp D 2 , D 3 , ••• Dn. Were we dealing with a person rather than with a hill, one tempting suggestion would be that each Dj is an embodiment of Dumyat. But it is not easy to see how to separate hills and their physical stuff in the way that some would say persons are distinct from their physical stuff. The trouble is that hills do not seem to have appropriate high-level properties, for example memories, intentions, beliefs, and ambitions. For a person to survive a change of body we might look for persistence of such high-level properties. And of course replacement of a person's body by another exactly similar body is indeed a change of body. The Dumyat case is thus like a change of body without persistence of the mind. A better option, perhaps, is to claim that each Dj is an extended stage of some same object in just the way that momentary stages of objects are stages of objects. But of what kind of object are we thinking? If we stick with the notion that our hill is a concrete continuant, then-unlike the usual situation in such cases-it fails to exhibit continuity among its stages. So have we here some special sort of continuity--continuity that allows little gaps? As we have already seen, there is a Hollywood sense of 'continuity' which is just right for the job here. But if we leave that sense of the term aside, we are left with the various kinds of continuities discussed in Chapter 2. We observed there, by looking at the tree and rocket examples, that some of the continuities exhibited by objects in their everyday careers are pretty weak. For instance, successive volumes occupied by a multi-stage rocket do not always show a high degree of overlap. But in the present case, unlike that of the rocket, we cannot even speak of weak continuity. There is in fact no overlap between the place occupied by any Dj just before a gap and the place of its successor in the very next 1~ of a second. For it has no successor then, and coincides with no place at all in that interval. As a desperate remedy perhaps we should allow the temporal jump, taking it as no more than a special case of the partial discontinuities already discussed. But this looks most oddas if we are trying to allow discontinuity as a special case of continuity. Some of the oddness of this course of action is perhaps reduced if we recall an ambiguity in the notion of continuity itself. For although Hirsch's conception is entirely natural if· we are thinking in terms of a continuous path traced by an item through space-time, not everyone has just this notion in mind when they use

Identity and Discontinuity


the term 'continuity'. Consider, for example, Parfit's, Lewis's, and others' accounts of psychological continuity. 7 None of these authors is unaware of the episodic nature of consciousness. Rather, what they mean by 'psychological continuity' is that each new mental episode goes on where its predecessor left off, whether or not any temporal gap separates the episodes in question. This is, again, our familiar Hollywood sense of 'continuity', and I do not intend to follow these authors in using the term in such an extended way. Rather, I will restrict the notion of continuity to what seems to be a perfectly natural conception even though, as we have already seen, we lack a satisfactory mathematical account of it. Using this very notion, then, it is clear that the Dumyat example is not a case of some rather weak continuity; rather, it is a case of something entirely different.


Perhaps we should consider assigning each Dj to Dumyat in much the way that each instalment of The Archers can be assigned to one and the same radio series. We have already noted that many sorts of episodic object seem to be non-concrete. But it would be simply question-begging to move from this observation to the conclusion that any discontinuous thing must be nonconcrete. So just what is the relation between discontinuity and concreteness? In dealing with functionally discontinuous things, 1 have taken it that after dismantling, the things persist, albeit in a somewhat scattered fashion and with some of their normal functions suspended. But some authors, like Hirsch and Nozick, find it natural to regard an item that is taken apart for repairs as going out of existence for a while and then coming back into existence upon reassembly. As already noted, not much would be gained by arguing about whether their description of such cases is preferable to mine. However, accepting their view for the moment, it is interesting to think about the conditions that have to be met for a newly assembled item to be numerically the same item as one previously disassembled. One such condition, drawn from what Quinton calls 'compositional theories', might be that the item in 7

See Parfit [1], [3], and David Lewis [3].


Identity and Discontinuity

question should contain the majority of the same components in much the same arrangement as in the original. 8 Another condition might be that some same sortal be true of both original and reassembled items. Still holding the matter of sortals in abeyance, it should be clear that the compositional criterion is of no help to us in the present case. Even if we suppose that a watch taken apart for repairs does go out of existence for a time, there is no suggestion that its parts could in turn undergo similar disassembly, and the components thus revealed undergo further disassembly ... and so on. For if such a process were carried on long enough we would have no components upon which a compositional criterion could plausibly operate. Molecules probably and atoms certainly would not yield identity conditions of the sort needed by the compositional theorist. The information that some object contains just the same atoms as some original object would not incline us, in the absence of other considerations, to count the two objects as the same. The compositionalist wishes to allow discontinuity of the watch along with continuity of its parts. So the identity of the parts themselvesor of a fairly significant proportion of them-is required to uncierpin the later identification of the repaired watch as the same as the original. Such an identity of components is just what we do not have in the hill case. Each Dj, of course, has much the same structure as its immediate predecessor (with due allowance for erosion, quarrying, and landslip). But whether the parts participating in the structure of a given Dj are the same as those in its immediate predecessor or successor is something that the hill's discontinuity prevents us deciding. So even the compositional theorist cannot come to our help here, and we are still faced with the problem of the relation between spatio-temporal continuity and concreteness. Let us then return to the conception of each Dj as an extended Dumyat-stage and think about the relation of momentary stages to the larger concrete continuants of which they are stages. In Quine's familiar example, each momentary stage of the Cayster is a concrete object, while the river itself is also 'a single concrete object extended in space and time' .9 On this account, momentary stages are parts of 8 9

Hirsch discusses this proposal in Hirsch [2], ch. 2. See also Quinton [2], p. 63. 'Identity, Ostension and Hypostasis', in Quine [1].

Identity and Discontinuity


continuants. Viewed in this way an object is (improperly) a part of itself. Moreover, there will be objects-an abundance of themmanifesting spatio-temporal discontinuity; it just happens that we, with our limited resources, lack the terminology, time, or patience for singling out such items. Thus Goodman deals with possible entities by speaking not of new, non-actual entities, but by speaking instead about actual, scattered, entities. The sum individual, he tells us, consisting of place p and time t (where p is a phenomenal visual field place not presented at t) 'misses being a place-time much as the scattered whole comprised of the body of one automobile and the chassis of another across the street misses being an automobile'.lo As has already been argued, a mereological stance like this can provide us with a way of saving the concreteness of an object while recognizing its discontinuity. Let us keep on the mereological path for a while to see if a solution is genuinely in prospect. All that we have established at the moment is that the various Dumyat-stages can be summed to make a single scattered individual, just as any other scattering of items can be said to constitute a mereological whole. But this in no way assures us that the mereological whole in question is really a hill. Perhaps the whole misses being a hill in the way that Goodman's scattered car components miss being a car. So we need to establish just what considerations would lead us to count a scattered whole as constituting an object belonging to some natural, artefactual, or other, kind. A comparison between the Dumyat case and a different form of discontinuity might help suggest an answer here. Figure 10 shows the history of a number of Dumyat-stages, whale-stages, and iceberg-stages, imagining this time that the stages are all of approximately the same length. II' 12 , and 13 are extended iceberg-stages which, let us suppose, we would unhesitatingly allocate to the same iceberg if they showed spatio-temporal continuity; and likewise for the whale-stages Wl' Wz> and W3 • The gaps after tl and t2 are meant to be very small, just as in the earlier case. So an observer watching location PI would seem to see Dumyat replaced instantaneously by a whale (at tl) and the whale by an iceberg (at tJ. Those imprellsed by spatio-temporal continuity as a criterion for identity and unaware of the brief discontinuities in lU

Goodman [I], p. 51.


Identity and Discontinuity

'------------- p pz Fig. 10.

time between Dp W2 , and 13 might conclude that some one thing could begin its career as a hill, then change into a whale before becoming an iceberg. It might be easier to take such a view if one believes, with Quinton, that 'the temporal parts of an enduring thing would have been a perfectly good thing of that kind if they had existed on their own without the other phases which in fact preceded and followed them', and then the subsequent discovery of the discontinuities might be viewed with some relief. 11 Now, in our latest case, DJ, D 2 , and D3 show spatial scatter as well as the son of temporal discontinuity already displayed in Fig. 10. The problem is whether there is any way in which a mereologist could allocate these stages to one single item, that item being a hill. There is of course no problem about allocating the stages to an individual. Our problem concerns the kind or sort (if any) to which the individual belongs. We might think that whereas in our previous case Dumyat was 'hopping' on the spot in the new case it 11

Quintan [2], p. 77.

Identity and Discontinuity


'hops' from place to place. But what could underpin this allocation of separate stages to the one object in either case? Our three conditions will do the job. Firstly, the structure condition requires that each stage and its successor be similar in geometric structure. The similarity does not need to be exact, but there are limits of a fairly stringent sort to the degree of structural change we would allow in cases where the stages are separated by such brief gaps. Our second condition requires similarity of material from one stage to the next, and again, although the requirement is not that there should be exact similarity, clearly we would not expect great changes to the material stuff of an object from moment to momentnot if we are going to count the object as persisting, anyway. Finally, the third condition requires that each stage--in Lewis's phrase-'causally depends for its character' on the stages immediately before it. 12 'Character' here means no more than structure together with matter. Applying the conditions gives us something like the desired result in both cases. Dumyat hops on the spot in the Fig. 10 case in that each stage seems to be causally dependent on the previous stage for its structure and matter, and the same relation between successive stages in the second case gives us grounds for allocating each of them, scattered as they are in space, to one and the same persisting object of a sort. The question of sortals will be discussed shortly. But that the persisting thing falls under the sortal 'hill' will, on my account, be a consequence of its manifesting appropriate structure and matter in each of its stages. We can further observe that the geometrical closeness in structure between stages explains the congruence, or near congruence, of the volumes of space coinciding with each stage.


It might be objected that there can hardly be a causal connection between successive Djs for the simple reason that there is no means by which any causal mechanism could operate. For no Dj is contiguous in space and time with its neighbour, and even Hume, at 12 Quoted from Lewis [3], although it should be noted that Lewis there is talking about what he wants in wanting his survival.


Identity and Discontinuity

least in some places, requires such contiguity between a cause and its effect (see Treatise I. iii. 15). Now, it can hardly be replied to this point that Dumyat simply goes missing in the brief intervals: for if Dumyat has popped into a higher dimension, or gone off elsewhere for the vital fraction of a second, we have no puzzle. We would then have no interruption in its existence and so no episodic object. And it might, of course, be argued that in the kind of case described here, it is far more sensible to think of Dumyat as having gone elsewhere for th~se brief intervals rather than to imagine it has gone out of existence. Likewise, physicists are reluctant to count any of the forces of nature as acting at a distance, although I know of no evidence so far to suggest that, for example, there aregravitons. The case of gravity is indeed illuminating here. For although gravitons would provide appropriate quanta for mediating gravitational forces, we are still quite clear that bodies have gravitational effects on other bodies even while doubting that gravitons exist. Perhaps, if we were really sure that Dumyat goes out of existence every so many hours, we would look for new quanta to explain the replication of changes wrought upon one Dj in its succeeding neighbour. And if there were not at least prima facie evidence for each Dj being causally dependent on its predecessor-if variation between a given Dj and its neighbours could be as great as we please--then of course there would be no case for claiming survival from one Dj to the next. Of course, as in Hume's famous discussion of the billiard balls (Enquiry, section IV), we are here speculating about things which threaten our underlying conceptions of physical reality. Principles of conservation of mass and energy, for example, are threatened by both kinds of case; if mountains did flash on and off, or if billiard balls ever remained at rest after colliding, then an overhaul of physics and mechanics would undoubtedly be required. This being so, it is natural to suggest that there really has to be more to the relation of cause and effect than is suggested by Hume's austere account, namely the constant conjunction of two kinds of event. The next step is to look for some mechanism which mediates the influence of one event on another. Anyone who is tempted to take such a step must then be doubtful that things like our hill episodes could exert genuine causal effects on each other, or-more precisely-be ingredients in events which are causally related. For what conceivable mechanism could bridge--as it were--the gap between the hill stages?

Identity and Discontinuity


In thinking about this issue, we come face to face with a much bigger dilemma posed by the tension between our need to find explanations on the one hand, and our use of descriptive models in science on the other. The dilemma is not easily resolved. To get its full flavour, we can consider the distinction between explanation on the one side and prediction on the other. In a classic treatment, Carl Hempel argued that the two notions are symmetric. As he put it: scientific explanation (of the deductive-nomological kind) differs from scientific prediction not in logical structure, but in certain pragmatic aspects. In one case, the event described in the conclusion is known to have occurred, and suitable statements of general law and particular fact are sought to account for it; in the other, the latter statements are given and the statement about the event in question is derived from them before the time of its presumptive occurrence. l ]

What Hempel means by 'deductive-nomological' explanation is one that satisfies the following pattern:

Covering Law ( s) Statement ( s) of initial conditions


The adiabatic lapse rate for dry air is 3 °C per thousand feet. The air sampled at point A is exactly one thousand feet higher than air sampled at B. Both samples are taken in the same non-humid air mass. The temperature of the air sampled at point A is 3 °C lower than of that sampled at point B.

Here the covering law---or laws-will enable the conclusion to be deduced logically from the statement of particular events-the socalled 'initial conditions'. The conclusion itself deals with further particular observations. In its crudest terms, then, the formal pattern of such explanation is Universal generalization: Specific initial condition: Conclusion (also specific):

Vx[Fx::J Gx] Fa Ga

Hempel's symmetry claim is that (with certain qualifications) the schema just given illuminates both the notions of explanation and prediction. Provided the covering generalization is a law---or, at 13

HempeJ [11, pp. 366-7.


Identity and Discontinuity

least, lawlike-then it, together with the initial conditions, explains the conclusion. 14 Think of observing that the temperature of the air sampled at two points differs by 3 qc. IfHempel is right, we explain this observation by appeal to the generalization and specific initial conditions. Alternatively, given the initial conditions then-with the help of the covering law-we can predict that the samples will differ by the stated amount. One difficulty with accepting the claimed symmetry is that we can apparently make sensible predictions even when ignorant of explanatory factors. Mendel, for example, was able to predict the outcome of cross-fertilization of different varieties of pea-plant while ignorant of what we would now take to be the explanatory causal mechanism-namely the genetic structure of the pea varieties in question. Likewise, epidemiological studies can give us the ability to make statistical predictions for disease rates in the presence of occupational or environmental hazards, even when we are at a loss to explain the mechanism underlying the disease, or are even ignorant of what the hazardous agent is. It is hardly a decent explanation of why a particular New Yorker has cancer of the lung to state that citizens of New York have high rates of lung cancer: indeed, it is not an explanation at all. What seems plausible, then, is that explanatory models will yield the ability to make predictions, but that predictive ability need not yield explanations. Hempel, and his defenders, can reply to this that the covering generalizations must not only be general truths, but must have the status of being lawlike. Yet we must be wary of building into the notion of lawlikeness any claim that law like general statements are bound to be explanatory. One interest in Hempel's account is that the schema he gives itself provides an analysis of the concept of explanation: the analysis thus cannot without circularity use the 14 Both Hempe1 and Goodman, among others, are aware of the problem involved in trying to specify just what makes a generalization a law or Jawlike. It is intuitively clear, though, that not every general statement can do the job done by a law statement. If all the coins in my pocket have milled edges, for example, it certainly does not follow that a certain 20p piece which is not now in my pocket would have had milled edges had it been in my pocket. So the universal claim that all the coins in my pocket have milled edges is not lawlike. However, the statement that all acids turn blue litmus paper red is lawlike; for it does follow, given the truth of that sentence, that if a cenain (non-acidic) sample ofliquid had been acid it would have turned litmus paper red. Classic discussions of this problem can be found in Goodman [1] and Hempel [1].

Identity and Discontinuity


concept itself. In the end, perhaps the fairest thing to say about Hempel's own concept of explanation is that it is a 'thin' one. Any account that insists on the symmetry he claims to hold between explanation and prediction will be an account of 'explanation' in only an attenuated sense of the word. Hume's austere account of causation is itself a precursor of Hempel's work on explanation. For Hume, at least some of the time, gives us an account of causality which seems to rob that notion of any explanatory role. As he puts it in section VII of the Enquiry: 'we may define a cause to be an object followed by another, and where all the objects similar to the first are followed by objects similar to the second. Or, in other words, where, if the first object had not been, the second never had existed.' Ignoring the worry that Hume should really have been talking about events rather than objects, we can see that this account deprives appeal to causality of any rich explanatory force. In the place of an appeal to powers, forces, or causal mechanisms, we have simply a statement of general correlation. There is thus great similarity between Hume's schema and Hempel's: Causal generalization: Initial condition: Conclusion:

All events of type A are followed by events of type B Event QC is of type A Event QC will be followed by an event of type B

It is hardly surprising, then, that this austere account of causation is just as unsatisfying as Hempel's model when our concern is with explanation. One natural way of trying to enrich the Humean conception of cause is by appeal to fine structural considerations. Running diamond over glass with sufficient pressure will cause a scratch in the glass. Why? The explanation, we might think, lies not in the generalization that past encounters between diamonds and glass have had similar upshots, but that the fine structure of diamond is such that it is harder than glass. Likewise, we explain why sugar dissolves in water, or why adding chromium to steel makes the resulting material less liable to rust, by appeal to features of fine structure. Here is an attractive move. Applied to Hempel's account, it suggests the analogous move of enriching his account of lawlike


Identity and Discontinuity

regularities by the addition of models. For Duhem such a move might be no more than evidence of shallow thinking. 15 But a suitable model for chemical reactions or for crystal structures will give us explanatory resources that generalizations alone cannot provide. This enriching move is not one to which I am at all hostile. On the contrary, we will shortly be looking at the relation between changes in things and alterations in their fine structure. For the moment, we can perhaps note that the Humean has two replies to this latest move. One is to point out that retreating to details of crystal lattices or chemical structure explains nothing; for at any level of structure we are confronted again with the same old causal questions. Why does this atom of sodium occupy a particular position in a salt crystal? Hume's reply would be that all previously observed sodium atoms in such environments have taken up similar positions. Thus we have not reached any kind of ultimate explanation by appealing to the crystal structure in an account of the behaviour of salt. A second reply is given by Hume himself. Suppose we say that diamond is hard because tetrahedral arrangements of carbon atoms are good at resisting pressure and deformation. But what justifies our confidence that the next diamond we pick up will have a similar fine structure? Only, apparently, a causal inference! As Hume says: The bread which I formerly ate nourished me; that is, a body of such sensible powers was, at that time, endued with such secret powers. But does it follow that other bread must also nourish me at another time, and that like sensible qualities must always be attended with like secret powers? The consequence seems nowise necessary. (Enquiry, section IV. ii)

Whereas the earlier Humean response threatened a regress of explaining constant conjunctions by underlying mechanisms which in turn display constant conjunctions explained by underlying mechanisms ... and so on, the second reply suggests that every appeal to an underlying mechanism is simply question-begging. These are undoubtedly powerful replies. And, like all the best forms of scepticism, it is hard to find convincing rebuttals. One consolation is that Hume's 'thin' notion of causality as constant 15 Duhem follows Pascal in making somewhat astonishing claims about kinds of mind: 'The English mind is clearly characterised by the ample use it imaginatively makes of concrete collections and by the meagre way in which it makes abstractions and generalisations. This peculiar type of mind produces a peculiar type of physical theory; the laws of the same group of phenomena are not coordinated in a logical system, but are represented by a model.' (Duhem (1], p. 86.)

Identity and Discontinuity


conjunction is ideally suited to our fantasy case. Once we are aware of the discontinuities afflicting hills, we can test to see if changes at one time leave their mark on later occasions. Sure enough, material quarried from one hill stage leaves its mark on later stages. As predicted by Hume, we will thus be led, having observed a change in one hill stage, to expect a comparable change in its successor. Of course, Hume would point out that no sound inference underlies this expectation. Hills that disobey the conservation principles are not logically impossible, and such things would not display this sort of regularity. One advantage of philosophical fantasy, however, is that we can control the plot, so long as we keep within the bounds of consistency. And in my story the hills obey conservation constraints of a sort: the stages do survive as their successors and we would, no doubt, in such situations find ourselves with suitably Humean expectations. It may seem that even the Humean has to be committed to spatiotemporal continuity as a necessary condition of identity by dint of one last set of considerations. According to these, since identity is a causal notion, then persistence through time is bound to involve continuity from one stage to another. It may be anxiety on this very point which led Shoemaker to write, 'if material persistence requires spatiotemporal continuity, then this is because the relevant sorts of causality require it'. 16 Suppose, for example, it were argued 16 The quotation is from Shoemaker [6], p. 242. It is interesting to note the similarities and differences between Shoemaker's account of unity as a causal relation, and my own. Many of his general remarks (e.g. about the inadequacy of the spatio-temporal continuity account of identity) are ones with which I have no quarrel. However, in his own account of how causation is involved with unity, there is a significant divergence from the position I argue here. Causation in my story is internal, so to speak, in the way it operates; it is a kind of binding force that ties the different components together into a long chain which constitutes the life history of 'he broad object. By contrast, Shoemaker gives an account in which causality has an ,xternal role to play. For him, a temporal pan is a complex of property instantiations. \ broad particular (like a saw) will thus have various causal powers in virtue of its temporal stages being co-instantiations of various properties. See his essays 'Causality and PropenJes' and 'Identity, Properties and Causality' in Shoemaker [6]. How can an account like this be relevant to cross-temporal unity? Shoemaker spells it out as follows: 'Since a specification of the essential nature of a property will involve a specification of the powers to which it has the potential for contributing, and since a specification of the powers will say what happens to their subject O'Ver time given certain conditions, the essential nature of a property incorporates the persistence conditions, that is the cross-temporal identity conditions, of the things to which it can belong.' (Shoemaker [6], p. 253.) As far as I can see, Shoemaker's account thus leaves to one side the kind of causal cormection among stages that are so significant to the survival issue.


Identity and Discontinuity

that if one event is the cause of a second event, then there is always some further event which is in causal relations to both. Likewise, if one stage is causally productive of some other stage, there would have to be some third stage which can be interpolated between the first two, and which stands in causal relations to both. The idea is that causation satisfies a kind of interpolation principle reminiscent of Craig's well-known lemma. A successful argument for such a causal principle would indeed show that the Dumyat case is not one where the various hill episodes can be causally related. Nothing in the points made so far about explanation count against this argument. And if the argument could be established, even a Humean would have to admit that causal relatedness implies continuity. However, I can see no reason for taking the argument seriously as a point about causation. Think of the spatial correlate of the argument. It would involve maintaining that between any two causally related parts there will always be a third part causally related to the other two. The closer we look into the fine structure of objects, the less plausible such a claim looks. Likewise, turning our attention in quite the opposite direction, it is clear that distant objects exert gravitational effects on the earth. The theory of gravity, however, in no way requires that we be able to construct a chain of interpolated bodies between the earth and such distant bodies. We thus come back to the case of gravity, which prompted the excursion into the theory of explanation in the first place. It now looks as if some ways of enriching the Humean account of cause would force us into denying that discrete and discontinuous stages are really in causal relations with each other. But we can also resist such enrichment by staying content with a 'thin' notion of cause (and with a correspondingly 'thin' notion of explanation). In doing philosophy of science, we sometimes fall prey to an all-or-nothing tendency, which I am eager to resist. The result of the present discussion should not encourage anyone to think that all cases of causal explanation will be equally thin. In some cases, we will be able to enrich the statement that there are Humean conjunctions with descriptions of mechanisms in terms of which we can understand the occurrence of the conjunctions. But there is no reason to believe this holds in all cases. We can state the principles governing gravity (for example, the inverse square law), associate gravity with a field, and still regard it as a force that acts at a

Identity and Discontinuity


distance. Our explanations of gravity may always remain 'thin', in my sense. Just so, if there turned out to be discontinuities of the sort imagined to afflict Dumyat, we would be able to give only 'thin' explanations of why changes in one stage bring about effects in later stages.


For the present I do not want to dwell at length on questions about matter, which will be clarified in the sixth chapter. We know that at the submicroscopic level what matter an item is made of is associated with the details of its fine structure. Our telemutation case in Chapter 3 has shown that things can survive a change of fine structure, hence a change of matter, so long as their macroscopic structure and overall dimensions remain fairly constant. Thus there would be circumstances where, if we managed to transmute the material of a certain object, while preserving its form, we would judge that the original object had, to some degree, survived in, or as, the new item. There are perfectly good reasons, given the nature of human perceptual mechanisms, just why we should make such judgements of survival. 17 It is hardly surprising, then, that the similarity of overall geometric structure between each Dj and its neighbours is an important component in underpinning our identification of the scattered whole consisting of the various stages with one thing of the sort hill. In terms of the notion of survival, the situation could be described in the following way. Each Dj survives as the subsequent D j + 1 provided our three conditions are met. And the chain of survival relations among the stages is precisely what establishes the claim that the scattered whole in question constitutes one hill-a claim that can be made in either the Fig. 9 or the Fig. 10 case. When we turn to the parallel psychological case in a later chapter, it might seem that at the level of functional description suggested by the use of terms like intention, need, desire, memory, and so on, a mental state is not to be taken as a concrete item. Nevertheless, it is possible to look for analogues for the three conditions. The contents of short- and long-term memory, material undergoing perceptual 11

This is argued; among other things, by McCabe in McCabe [1].


Identity and Discontinuity

processing, one's store of ambitions, intentions, and so on can call be regarded as in some sense psychological matter. This material is no doubt structured within devices (thus long-term memory has, among others, a semantic level of organization) which themselves have structural relations with each other. The recollecting from long-term memory of an experience entered some time before would be an example of how a structural change in one state is in part causally dependent on a material (and associated structural) change in an earlier state. We will follow up to implications of this analogy when we come to discuss the puzzles associated with amnesia and branching in Chapter 9. To suggest a different application of the model proposed here, let us suppose that, as a kind of practical joke, the Stirling Philosophy Department records an entirely bogus episode of The Archers, and cunningly arranges for it to be broadcast in place of one of the regularly scheduled episodes. A soap opera is, of course, an episodic object whose scattered episodes are all allocated to the one item. Now it is immediately clear that our bogus episode of The Archers could pass for the real thing if we ensure that it satisfies the appropriate correlates of our three conditions. It might be objected that matter, structure, and lawful causal relations can only be talked about in a metaphorical way in connection with such an example. But if we achieve the dovetailing of our bogus episode with its neighbours, by means of reasonably stable characterization, good mimicry of the voices of the several characters, continuation of the action, adoption ofthe appropriate style, and so on, then we would have again satisfied at least metaphorical counterparts of our three conditions. One way of thinking about it is as follows. Let the fictional matter of an episode be the various characters, the village of Ambridge, its surrounding farms, and a number of happenings (which we can take to be event types). The structure of an episode is then given by the various relations exhibited in the episode among the characters, places, and happenings. On this view, structure changes throughout an episode. The supposed causal dependence of one episode on another may be revealed in the way the terminal structure of one episode induces a specific initial structure in its successor. More likely, perhaps, structural changes in earlier episodes map onto the structure of the current episode in an appropriate (though complex) way. Indeed, if we carried it off well enough, our episode might well have to be reckoned to the total of

Identity and Discontinuity


episodes that constitutes The Archers, and in this way our example threatens the view that the identity of the series depends on the authorized script in much the way that the identity of a musical work might be held to depend on the composer's score. IS We can further test the three conditions by considering what our response to various other cases might naturally be. Would we be inclined to count a stage of Mount Everest along with a subsequent stage of the Empire State Building as constituting one thing of a kind? Surely not, and the reason is that we have neither a structural, material, nor causal basis for such a claim. Now visualize a corridor with a number of similar doors facing on to it. We discover, let us suppose, that doors, too, are afflicted with Dumyat-like discontinuities. How would we decide whether a given door was hopping on the spot or changing places with others elsewhere on the corridor? Our conditions suggest, once again, the way to seek an answer. We make a scratch on one door-stage and then see if this slight structural modification reappears on the next door-stage to occupy the same doorway (assuming, of course, that we have settled the matter of the identity of the doorway). If it does, and the new doorstage is made of similar material to the original, then we have a fairly good case for concluding that we can allocate the new stage to the same door as we allocated the original. Our scratch test also provides prima facie evidence of the right sort of causal dependence between stages.


All this looks like good news for those who are happy to accept a mereological stance. Since they are comfortable in the recognition of scattered objects as individuals in any case, mereologists would find the suggested conditions helpful in letting them identify, from among the vast range of indi viduals, those which are unitary, albeit discontinuous, objects of a kind. Would they need, in addition, a sortal condition to justify such assignments of stages to scattered wholes? We might at first think that it is the fact that each Dj falls under the same sortal which along with our other conditions clinches the question of Dumyat's identity. But here we must be 18

See the discussion in Goodman [3].


Identity and Discontinuity

wary; for it is not clear what explanatory force such a sortal condition would have. In the case of living things, what would lead us to class a number of objects as all falling under some same sortal? Domestic cats all display certain structural similarities to each other, have similar material constitution and depend causally for their structure and matter on the structure and matter of their parents. Indeed, as I have already argued, one natural way of thinking about survival is in terms of the way parents can be said to survive in their offspring, and where we have survival then we have some degree of structural similarity, lawful causal dependence, and-often-similarity of material constitution. So if-for the sake of the argument at least-we suppose that the sortal 'hill' can apply to any of our stages in the Dumyat case, our three conditions seem to explain why the same sortal applies to other stages. We add nothing to the conditions by attempting to impose a further requirement concerning sortals. But does the sortal 'hill' apply to any of our stages? This is a matter over which we must be extremely careful. To count the various scattered Dumyat stages as stages of one hill is not to imply that any of the stages themselves is, properly, an item of the sort hill. Quinton's doctrine might come to mind here. No doubt there are things such that ten-second, ten-hour, or ten-day manifestations of them would be, in his words, 'perfectly good things of that kind'. But it is not clear that whales, icebergs, and hills are examples of such things. Suppose something looking like a mature whale puts in a ten-minute appearance in my garden. Naturally, I wonder where it came from and where it subsequently wept. Now suppose that I find it had no previous or subsequent history: I saw the entire life of a ten-minute individual. Was it a whale, or a stage of a whale, that I saw? Either proposal seems doubtful, and the example suggests that Quinton's claim is not at all self-evident. To accept it is to deny that persisting things have histories-where this means more than a life of ten minutes-though of course this claim itself is not safe from scepticism. 19 There is, however, a truth which corresponds to Quinton's claim. This is that, roughly speaking, each stage allocated to an item of kind S should possess that structure in virtue of which a succession of such stages, connected according to our require19 No claim, after all, is safe from scepticism and we can easily make sense of the last-minute creation puzzle. We do not know that the universe did not come into existence only five minutes ago, including us and all our apparent memories.

Identity and Discontinuity


ments, would be an S rather than anything else. We will give this issue some more attention in the next chapter. This, then, completes the description of one kind of solution to the puzzle we started with. To accept the solution requires in turn an acceptance of a mereological stance, and this is certainly not congenial to everyone. However, the mereologist has proved to be in a strong position for making sense of the bizarre state of affairs in which we find ourselves confronted with a world of discontinuous objects. What we must do now is shift metaphysical focus for a while and consider the prospects for a non-mereological solution. Such a solution is perfectly feasible, though leaving us with a rather more fragmented world than the mereologist's. For I want to suggest there is at least one way of coming to terms with the Dumyat puzzle which allows that in such an eventuality we would have to recognize that the world contains no broad objects. Whereas the mereologist saves the broad object at the cost of permitting it to be discontinuous, the alternative about to be mooted recognizes that each Dj is a new object and tries to preserve our use of the same name for each such object in the face of this recognition. Let us agree that considerations about structure and so forth do indeed suggest that each Dj survives as its successor. Why not now use the name 'Dumyat' for any extended stage and use the argument from survival to justify using the name for any appropriately related stage? Of course, we are not now, literally, speaking of stages at all, for there is no broad object to which we are assigning the various things called 'Dumyat'. Rather, we are following the convention that we use a name to apply to an item or to any clear survivors or ancestors of that item. This is to depart from the orthodoxy that the identity of an object is what makes sense of the application to it of the same name at different times. But in a world lacking suitably broad objects, there may be little point in trying to persist with the orthodox view. Strictly, then, any claim about Dumyat is really a claim about past, present, or future Dumyats. For any item called 'Dumyat' will-if our original case is recalled-last no more than 167 hours; and of course, we will not know in the normal run of things whether a new Dumyat has come on the scene even as we speak about the hill. Can we really make sense of this as an alternative to the mereological view just canvassed? I think we probably can, once we have managed to focus down onto a world where things have strictly


Identity and Discontinuity

limited persistence. Nor, incidentally, is it being suggested that in some way Dumyat supervenes on the various short-lived stages occupying a certain place. Rather, the present suggestion is that we drop the notion that there is some one persisting thing that is the aggegate of all the structurally and materially similar items that successively occupy one and the same place and that we do not retreat either to a conception of some supervenient item occupying that place. Instead, we believe that there is a sequence of Dumyat's, each one a different hill, and the survival of each hill in the sequence as its successor is what legitimizes the application of the same name to each succeeding item. We are in a position rather similar to that of the person who ostends various river-stages while uttering the same name; but instead of this procedure being used, as in Quine's example, to show that we are talking about one spatio-temporally extended object, we are to be taken as indicating that we are talking about a sequence of items, each of which is a survivor fo the item originally ostended. One argument which is not being suggested here would be based on the theory that some underlying causal process is responsible for the appearance of each succeeding hill. We are thus not taking seriously the possibility that Dumyats come about as the result of a production process. Of course, it is open to us to suggest that there is some, perhaps divine, production-line involved in churning out each of the separate hills. But, as we saw in our discussion earlier, although each Dumyat might then replicate the others (plus or minus quarried stuff and forestry plantations) we would not be able to talk of one surviving in or as any of the others. We would have successive tokens of the type hill, indeed, but no survival. That the production story is not what is being given here is quite easily seen, and since it is different it will require a different treatment. In telling our story, it was made clear that changes in one episode showed up in the subsequent ones. So the effects of today's quarrying, for example, will still show uP. next week and the week after that. This is typical of the situation in a copying process. Even if a divine being were producing each subsequent hill episode, the previous episode still has a causal role as the model from which the copy is made. So the theory of survival thus remains applicable, and our suggestion is that the survival of each short hill as its sucessor would legitimize our application of the same name to admittedly distinct objects.

Identity and Discontinuity


Now although such a procedure seems possible in theory, would it not prove entirely unworkable in practice, due simply to the enormous difficulties involved in saying simple things like 'I mean to climb Dumyat tomorrow'? I suspect that, if we were to adopt this second solution, then, whatever our .official metaphysical view, we would find ourselves speaking about Dumyat in much the way we currently do. This is not all that hard to explain. Mter all, we do use language in other contexts that seems natural even while we are aware that it is not quite exact. For example, the sentence 'The sun rose at 8 a.m. this morning' clearly needs to be paraphrased in terms of one mentioning the rotation of the earth before a truth can, strictly speaking, be expressed. A worse problem may seem to confront us if we recall the worries already expressed about Quinton's doctrine. Would not the claim 'Dumyat is a hill' now fall foul of the criticism that what we are doing is predicating' ... is a hill' of some temporal phase of an object? This worry, though, is seen to be misplaced once we recognize that, on the view presently being canvassed, there is no one thing which each Dumyat is. Some objects may run for more than 167 hours. But in a world of impermanent things, perhaps 167 hours of a hill is all that we are going to get of it. The broad object, then, is the 167-hour hill; thus, while the mereological response to the puzzle was to wonder what features ensured that each Dj was a temporal part of the same hill, the present view takes these same features as explanatory of why we give the same name to each member of a sequence of distinct hills. It does seem to me that this latest solution is considerably less attractive than the mereological one. Indeed, I have pursued it mainly to be able to contrast it with the mereological one, for this contrast reflects two different ways that science can have an impact on our thinking. According to our latest 'solution', we go on speaking as before while recognizing that our discovery has revealed a gap between our natural way of speaking and the way we have come to think-under the impact of science-things in fact are. The mereological solution, on the other hand, can be compared with the scientific discovery that the best microscopic description of seemingly solid and continuous objects requires us to recognize that they are composed of spatially discontinuous aggregates of tiny parts. The hardness of diamond, for example, can be explained by the structural relations among its discontinuous parts; likewise, the structural, and other, affinities between stages explains, on the


Identity and Discontinuity

mereological conception, how the scattered Dumyat-stages make one hill.




The three conditions appear to be sufficient rather than necessary for the survival of one stage as another. Given the plausibility of the mereological account, we might wonder if survival always underpins identity. One problem with taking such a line is that we can easily imagine a situation in which, after one of the gaps, two hills appear instead of just one. Indeed, these may be 'identical', and of course, this would mean that Dumyat was subject to fission. We have already looked at a case of fission, namely, the peculiar case of Theseus' ship (see Fig. 11). This time, we have changed the labels on the diagram, and have also acknowledged explicitly that we are thinking of three possible worlds, Wp W 2 , and w 3 • Our symbols 'a', 'b', 'b" and so on, this time are used, in Kripke's sense, as rigid designators for broad objects. 2O Each constant thus designates the same object in every world in which it designates anything at all. Now, a plausible view of the Theseus problem is that in wj ' b is a continuer of a in Nozick's sense, while in Wp c is a continuer of a. As he puts it: 'To say that something is a continuer of x is not merely to say its properties are qualitatively the same as x's .... Rather, it is to

............. b'

•••••• b

.. .......





..- c'

... .........



Fig. 11. 20 See Kripke (2]. Notice that I am not here declaring myself for the necessity of identity, or for any associated doctrine of essentialism. Rather, I use the Kripkean mechanism simply to engage with those whose objections to the 'only x and y' principle start from Kripkean premisses. Essentialism and the necessity of origin are discussed funher in Ch. 5.

Identity and Discontinuity


say they grow out of x's properties, are causally produced by them, are to be explained by x's earlier having had its properties and so forth.,21 Using Nozick's terminology, the puzzle in w 2 is to determine which of b and c is a's closest continuer, for in his view 'to be something later is to be its closest continuer'. An earlier proposal that we have already examined, due to Hirsch, is that we look for a spatio-temporally and qualitatively continuous succession of object stages that minimizes change. If we allow that there is continuity of material things in w 1 ' we can see that both the change-minimizing and the closest continuer proposals might seem to favour the verdict that a is one and the same as bl, although neither Hirsch nor Nozick try to force this conclusion. The theory of survival might seem to be usefully supplemented by these accounts. The theory allows that there is survival in each of wl' W 2 ' and W3 of a, allowing that in W 2 ' a survives as both b and c. Why not, then, let change-minimizing or closest continuerconstraints determine our choice between the rivals in W 2 and thus settle the identity question? We could even introduce the notion of the closest survivor to apply to the item for which we would give our identity verdict. As we have seen, one virtue of thus speaking would be that we do not get distracted by considerations about spatiotemporal continuity. Unfortunately, there is an objection to this way of proceeding. For the proposed way of dealing with identity in the w 2 case seems to leave us facing an uncomfortable dilemma. Either it turns out that what thing something is depends on the existence of other things, or the very existence of something depends on the existence or nonexistence of other things. We can use an example from personal identity to illustrate the first horn of the dilemma. Suppose, as in Bernard Williams's example, I wake up one morning with all the memories of Guy Fawkes. 22 As we will see later, some theorists regard the presence of memories as a sufficient condition of personal identity. But if we agreed, even just for the argument's sake, that I am therefore Guy Fawkes, we get into trouble. For tomorrow,You may wake up with all Guy Fawkes's memories. By the memory criterion it follows that you are Guy Fawkes. But you and I are two l


21 22

Nozick [I], p. 35. The example is from Williams [I].


Identity and Discontinuity

different persons. So either Guy Fawkes is two different persons, or neither of us is Guy Fawkes. The problem this argument reveals is this. Surely, if I am Guy Fawkes, then the appearance of a rival cannot make any difference to this truth. So, in general, if x is one and the same asy, the truth of this cannot depend on facts about any other things. If we call this the 'only x and y' principle, then it seems to be incompatible with the approach just proposed-an approach that could be called a 'best candidate' theory. As Wiggins puts it, if we are willing to give our vote to a's closest survivor in w Z' 'we could walk up to the antiquarian's relic, seen as a candidate to be Theseus' ship, and say that, but for the existence of its rival ... it would have veritably coincided as a ship with Theseus' original ship. But the idea that in that case it would have been Theseus' very ship seems to be absurd.'z3 What Wiggins is arguing here is that if we take cc' and 'c" as names of one and the same ship, then it cannot be the case that this ship has the same life history in both W z and in w 3 • In w 3 ' if c' = a then c' comes into existence at the very same time as a does. But this cannot be true of c in w z. But those who believe in the necessity of self-identity will see the best-candidate theory as denying this principle, a principle enshrined in Butler's maxim that each thing is what it is and not another thing. 24 Wiggins adds: 'There is a temptation to add as a step to this argument: nothing might have been a different entity from the entity it actually is. But the temptation is to be resisted. We are discovering in this argument ... the real intuitive grounds for doubting that anything might have been a numerically different thing from the one it actually is.'25 Recently, Harold Noonan has tried to strengthen Wiggins's argument here, thereby supplying the other horn of the dilemma. For, if each thing is necessarily what it is, then what it is cannot depend on the existence or non-existence of rivals. So, then, let us suppose that we avoid this particular horn of the dilemma by allowing that c is distinct from c'. The four-dimensionalist can take just such a line. But to do so is to risk being impaled on the other Wiggins (9), p. 95. This remark has become a philosophical commonplace since Moore used it as the epigraph for his Principia Ethica. However, despite the vast number of authors who cite it, the source of the remark in Butler is seldom given. After a vain search through some of Butler's works, I have decided to maintain this tradition. 25 Wiggins, loc. cit. 23 24

Identity and Discontinuity


horn: for it now seems that c' is not present at all in w 3 • So what we are calling' c" in W 2 would not have existed at all but for the presence of its rival, b'. But this is just as absurd, Noonan claims, as our previous position. Now what is this thing that would not have existed at all but for the existence of b'? Noonan speaks of it as 'the plank-hoarder's ship in situation 2', and perhaps the worry here could be expressed in the following way. In W z there are two ships, namely a (Theseus' original ship) which conventional wisdom identifies with b' (the continuously repaired ship), and also c (the plank-hoarder's reconstruction). But in W3 there is just the one ship, a, which the best-candidate theory identifies with c', the plank-hoarder's one. The plank-hoarder's ship, c, is a ship in its own right in w z' while in W3 it is but a temporal part of a much broader entity. If we held the view that temporal parts of ships, or aggregates of such parts, are not themselves ships, then our present difficulty would seem to be intensified. For whereas c in w 2 is a ship in its own right, there is no such ship present in W3 (for in that situation the plank-hoarder's ship is said by the best-candidate theory to be none other than Theseus' original ship). Noonan, however, cannot claim this kind of support for his argument. For in another recent paper he explicitly maintains that 'it is reasonable to say that (some of) the temporal parts of an enduring object of sort S are themselves of sort S', and the qualification 'some of' here simply rules out those stages that are very short. 26 Let us suppose-for the sake of definiteness-that in w 2 ' the plank-hoarder's ship comes into existence at the moment the hull is complete. On Noonan's own view, then, an entity of the sort ship must also come into existence at that time in w 3 • So some same ship is present in both w 2 and W 3 , although only in W z is that ship a maximal aggregate of ship stages. So c is present both in w 2 and in W3 (although in W3 it is only part of the broader ship c'). This, however, undercuts Noonan's attempt to convict his opponents of absurdity. As he writes, the best candidate theorist' ... must then acknowledge that the following is a possibility: we could walk up to the plank-hoarder's ship in situation 2 ... and say (truly) that, but for 2. In Noonan [7]. To be fair to Noonan, he does not give any support in the paper to the temporal-part metaphysic. Rather, he argues for the plausibility of the claim given provided we allow that objects have temporal parts.


Identity and Discontinuity

the existence of its rival it would never have existed at all. >27 So far from this being a claim to which the best-candidate theory is absurdly committed, it is one that-by Noonan's own lights-it would have to deny. It is helpful at this point to consider what a mereologist would be able to say about our problem. Using the term individual in the mereologist's technical sense to apply to any object however sCattered it may be in space-time, if a temporal part of a ship is an individual, so will a collection of such parts be. Further, the natural way for a mereoiogist to read the labels in Fig. 11 is as rigid designators for collections of temporal parts (Nozick, for example, does so in his discussion of related problems in the notes to his book). On this reading, of course, b = b' and c = c', quite uncontroversially. The best-candidate theorist who is also a mereologist can now make identifications as follows. In w1 Theseus' ship is the individual consisting of the collection a together with the collection b (let us write 'a + b' for short). In w 2 Theseus' ship is also a + b (notice that a + b = a + b' ) and in W3 Theseus ship is a + c. Of course, the individual a + c (which is none other than a + c') is also present in w 2 ' but in that world, it is not a ship, let alone Theseus' ship. Noonan is not out to get the mereologist. On the contrary, he explicitly allows best-candidate theorists to take the four-dim ensionalist view if they so wish. But can he now convict me of absurdity in any of the things I have just said? Perhaps the point is that the object a + c is just not present in w 2 at all. Since the stage a is, so to speak, already allocated, along with b to Theseus' ship, then there is nothing, namely a + c, which might have been a ship in other circumstances (say, those which prevail in w 3 ). But why deny the existence of the mereological individual in this way? One motive for doing so would be the belief that distinct objects cannot have temporal parts in common. Now, in the spatial case, there is no puzzle. The sides of my desk and the top of my neighbour's desk might, in other circumstances, have constituted one desk. Of course, they do not do so in this world. So the same parts can be parts of distinct objects. In the temporal case, we can think again of

27 In Noonan [6]. The arguments attributed to him in the text are also presented in Noonan [5], [7], and [8].

Identity and Discontinuity


Hirsch's example of the tree and its trunk. 28 A tree is not the same thing as its trunk. But one year, let us suppose, the tree is pollarded so that every branch is cut right back to the trunk. Until new shoots appear, the tree and the trunk coincide. Later, the tree recovers: it and the trunk continue to have different life histories, but these histories now share a common stage. Likewise, a + b' and a + c, though distinct, share a common stage. So it looks as though the best-candidate theorist who opts for the four-dimensional way can deal with the fission case without falling into absurdity after all. It does follow from this, certainly, that one and the same thing-Theseus' ship-is identified with different mereological individuals in different worlds, namely a + b' and a + c (remember that we are still using 'a', 'b', and so on, as rigid designators for collections of stages). This, though, is hardly surprising, given that principles of identity do not apply to cases of transworfd Identification without some modification: For alternative worlds are precisely those in which the possibly different histories of objects are realized. On what I have called the 'conventional' view, Theseus' ship is present in all three depicted worlds. It has the property of being a + b' in w 2 and of being a + c in w 3 , and with this kind of world-indexing there is no problem about the necessity of such claims as 'Theseus' ship = a + b". Without world-indexing of properties--or some equivalent device---we would have probtems, and not just problems about the necessity of identity. So we could now perhaps try to give a mereological gloss on Butler. Each individual is the individual it is and not another individual. But what sort of thing an individual is in one world (if any) depends on the existence or non-existence of other individuals in that world. Salmon, as pointed out in the footnotes, has an argument rather similar to the one I give here, only his involves appeal to the matter out of which ship a is constituted. His account, like mine, preserves the necessity of identity. Thus it is not a contingent fact, for his account, that Theseus' ship is built originally out of certain planks and other bits. But it is a contingent matter thereafter just what matter we deem as constituting that particular ship. More will be 28 Hirsch [2], ch. 2. Salmon does the same trick using matter that I perform here using stages. He notes that it is a contingent fact which matter constitutes a certain ship at a time later than its original construction. See Salmon [1], Appendix 1.


Identity and Discontinuity

said about problems of necessity of origin and identification of items across worlds in the next chapter.


More will also be said in the following chapter about transtemporal and crossworld identification. It is already clear from my response to Noonan's arguments in the previous section, however, that the analogies between worlds and times are likely to be worth pursuing. Just as the same item has a number of different temporal parts, so the same item will be constituted of different stage collections in different worlds. It is worth, however, noticing just what elements of my treatment of the best-candidate issue have involved ad hominem attacks on Noonan, and which have involved a real rejection of the 'only x and y' principle. Although Noonan's is the most sustained attack on best-candidate theories that I know of, it would be a breach of my own methodological norms to argue that my attack leaves it unnecessary to argue separately for the falsity of the 'only x and y' principle. There are two claims which Noonan deems to be absurd but which, on my account, are sensible commitments of those willing to adopt the four-dimensional account. These are: (a) events which constitute the origin of some entity of a certain son in one situation may not constitute the origin of that, or any entity of the kind, in a second situation, even though all the events constituting the history of that entity in the first situation remain present in the second; (b) two events may be pan of the history of a single entity in some situation, but may fail to be part of the history of that, or any single entity of the kind, in a second situation in which both they, and all the events which were parts of the history of the entity in the first situation, remain present. 29

Now Noonan does not take seriously the thought that the second and third situations depicted in Fig. 11 are likely to differ in ways consequent upon the premature destruction of the original ship in the third case. Of course, in any real situation where a plankhoarder was at work, the history of the hoarder's reconstruction is 29

Noonan [6], p. 81.

Identity and Discontinuity


quite likely to be influenced by the early demise of the continuously functioning ship. However, even if we stick with the artificial restriction that the two situations are alike in all respects except for the early destruction of the continuously functioning ship, we can see that there is nothing absurd about either (a) or (b). What removes any puzzlement we might at first feel in the face of these claims is the distinction between mereological individuals and objects belonging to natural or artefactual sorts or kinds. It is true that all the eVarises even when we take brief cross-sections of our experience. Here I am sitting at a desk, writing, thinking, occasionally looking at the furniture in my room or the trees outside my window, feeling slightly disgruntled, aware of a desire for a cup of tea, and so on. All these things are part of my psychologial experience here and now, and so they are one and all mine. But what gives this bundle of disparate items its unity? By contrast, there is the diachronic unity problem. This current bundle of Humean 'perceptions' has multiple relationships with past such bundles. So not only have we the problem of explaining what makes all my current psychological states mine, but we need to come up with some account of how these past bundles-in some cases very different ones-have been mine as well. And some past bundles which were very like this present bundle have not been mine at all. As we will see in the next chapter, the theory of matter, structure and cause can go some way towards solving both problems. My current psychological bundle has a certain structure among its elements, even though I am not consciously aware of all the ingredients that participate in this structure. Thus my processing mechanisms, and many aspects of my cognitive equipment, function in a way that is quite hidden from conscious inspection. Likewise, successive bundles that are mine stand in the survival relation to each other. Again, it should not come as a surprise that some of these items related by survival are not themselves available to self-consciousness. This, however, does not diminish their importance in the account of what constitutes a unified life. Historically, the most sophisticated response to Hume came from Kant, who attempted to answer both unity problems. My ability to fuse memory, present sensory awareness, and central conceptual processing-an ability essential to my recognition of my desk and my application to it of the appropriate concepts-shows that one knowing mechanism, one centre of consciousness, is constituted by these various strands in the bundle. Moreover, my recognition of objects as persisting over time is itself a simultaneous celebration of the persistence through time of this one processing bundle, namely me. Notice how clever Kant's move is. For bodies, indeed bodies


Memories, Bodies, and Survival

other than my own, have been brought in on the act in order to solve both the synchronic and the diachronic problems. 23 Kant's appeal to the objects around m33 But now what is being denied is that we have any 33 Here is Strawson's remark in context: 'What we have to acknowledge ... is the primitiveness of the concept of a person. What I mean by the concept of a person is the concept of a type of entity such that both predicates ascribing states of


Memories, Bodies, and Survival

concept or idea of a primitive sort of a pure subject of experience, of a pure Cartesian ego (in the popular, rather than the scholarly, sense of that notion). Thesepoints about concepts, it seems to me, leave plenty of scope for rival theories about the nature of things falling under the concept in question. From the fact that a certain recognitional capacity is primitive and that certain items are basic in our scheme of recognition, it does not follow that there is nothing to be said about the identity of the recognized things. So what sense, then, can be made of no-ownership theories? Just this, that whatever the situation in regard to our recognitional capacities, the following questions are open and worthy of consideration. First, how much unity do I possess at this moment? Reconstrued in terms of my bundle, this question asks to what extent my various mental states and dispositions are co-personal, that is correctly attributed to one and only one subject of experience. That there is a question over the existence of such synchronic unity is shown by the fact that in cases of multiple personality, for example, there is no guarantee that we would not wish to develop an account which permits more than one person to be associated with a particular body at a given time. The second question which is also open, and not settled by the considerations given by Strawson, concerns the degree to which I remain one person over the various vicissitudes of my history. Of course, the pedantic claim here is that the question is not open, for the reference to 'my history' already reveals the answer. But it is apparently sensible to wonder whether certain changes in someone's character are enough to yield a new person. Encouraged by these reflections, some-like Parfit-do not balk at writing of their former and their future selves. If it is agreed that there are open questions here, then it has to be conceded that there is something in the no-ownership theory after all. For in each case, the interesting question has concerned the number of owners of a set of mental experiences. consciousness and predicates ascribing corporeal characteristics, a physical situation etc., are equally applicable to a single individual of that single type ... a necessary condition of states of consciousness being ascribed at all is that they should be ascribed to the very same rhings as certain corporeal characteristics, a certain physical situation, etc. That is to say, states of consciousness could not be ascribed at ail, unless they were ascribed to persons, in the sense I have claimed for this word.' (Strawson [I], pp. 101-2.)

Memories, Bodies, and Survival


To talk in tenns of bundles, however, is not to adopt the view that persons are ontological parasites in Chisholm's sense. As we have already seen, the ontology of particulars is complicated by the fact that the very same particular can be considered relative to various dimensions. This consideration itself shows that simple reductionism with respect to any particular is not plausible. The very same thing is present in a certain volume of space, at a number of different times, and in a number of different worlds. None the less, the spatial unity of a particular consists in a certain sum of causally interdependent parts, its spatio-temporal unity consists in survival relations of the appropriate kind among its stages, and its cosmic unity consists, according to the theory of real possibility, in the similarity among the sums of its temporal parts in different worlds. Anyone accepting this kind of view about particulars is going to accept the same account applied to persons, and acceptance of it is incompatible with straightforward reductionism.


This quick sketch of some central issues concerning persons and personal identity gives a background against which I can now put forward my central suggestion. There is, I suggest, no well-defined concept of a person which can be analysed and put forward as the concept with which we all operate. Nor is it clear, in my view, that talk about persons, consciousness, self-consciousness, and the identity of animals is even consistent. In some circumstances, I would conjecture, we are likely to be body-minded, as when we discuss the identity of amnesics. More will be said about amnesia in the following chapter. By contrast, we are able to switch to psychological accounts of personal unity in order to discuss (even sceptically) Lockean soul !Iligration and the like. Our everyday modes of thinking about persons, that is, are hospitable to both body-based and memory-based accounts of identity. An intennediate strategy, which perhaps commends itself to us strongly in these days of materialist and functionalist theories of the mind, is that persons are to be individuated by a brain-based theory of identity. This strategy seems to hold together both physical and psychological criteria of identity, and we will study it further in the following chapters.


Memories, Bodies, and Survival

As we have also seen, Parfit has a very interesting argument to show that personal identity is not what matters. If he is right, then there is a particular difficulty we have to face in talking about our concept of the person. We cannot put the problem of my survival in terms of my identity with some future person; nor can we even use the possessive 'my' with any assurance in talking about 'my survival'. It must be borne in mind therefore that to talk about Parfit's concept of the person is already to describe his position in a misleading way. But-given our current linguistic resources-it is convenient to continue talking in this way. For it is important to recognize that he has made a distinctive contribution to the debate on the nature of persons. If my overall approach is along the right lines, then we will have scope for developing a range of accounts of personhood, including Parfit's. The indeterminacy in our everyday ways of thinking, speaking, and worrying about persons shows up most clearly in our ability to sympathize with, and feel attracted to, very different accounts of personhood. The puzzle cases confirm, as it seems to me, the fact that we have no very coherent concept of a person underlying our everyday talk and behaviour. If we do not take the Parfit line, then the most attractive alternatives are either a brainbased theory of personal unity or an account that starts from the position that personal identity is primitive or basic. I have already given reasons for doubts concerning this last position. We are left, then, with two promising models for developing the concept of a person, each with distinctive problems. Given the general approach urged earlier to problems of identity, it is hardly surprising that I should tackle the issue of personal identity by thinking of us as structured Humean bundles of perceptions, experiences, and the like. After all, any account of bodily identity would, in my view, have to give an account of the survival of body stages, and this would involve discussing the structure, matter, and causal connections among body stages. It is perhaps surprising that those who might be seen as modem successors of Locke--Parfit and Wiggins, for example--have found themselves in difficulty when dealing with even so straightforward a case as amnesia. However, attention to the issue of psychological structure and matter will give us the means in the next chapter to avoid some of the difficulties these theorists have made for themselves. Again, the simple distinction between production and

Memories, Bodies, and Survival


copying processes sheds light on some of the issues that baffle us when we think about persons. This talk of developing a range of concepts of the person must not, however, lead us to believe that we will get to a correct answer in the end about the nature of persons. There may be occasions when conceptual development can lead to our selecting one developed concept as being, for certain purposes, the best one to use. But I will not be making this claim for the concepts developed here. If this result is disappointing, it is at least not depressing. For any progress that is made in getting clear about the issues involved in conceptual development and about the range of concepts open to us for deployment is dependent on the prior acceptance of the case I am making. It will be sufficient, then, if I can at least convince the sceptical reader that there are alternative concepts of the person, without going on to argue the merits of one particular development over any other. One further peculiarity of my treatment should be noted. As is well known, mental phenomena are not well understood: we have no satisfactory, generally agreed, theory of mind. Further, the whole question of content is puzzling, whether we are thinking about the content of sentences or the content of mental states. Even those who deny dualism may therefore be reluctant to do what I do: this is to speak in causal tenns about psychological states and their relationship to other psychological states and to physical things. Causality itself is an odd notion, and certainly not one about which we can be ultimately very comfortable. However, it is undeniably useful, and it seems to have a use for us here in providing a convenient way of regimenting, for example, the discussion of amnesia that follows.


Amnesia and Other Problems 9.1. CONNECTED NESS AND CONTINUITY

As I have suggested, Parfit, like other neo-Lockeans, may be regarded as developing a concept of the person. It will not be argued that this concept is a better development than its nearest rival. What can be shown, however, is that there is the possibility of developing a concept like Parfit's in a consistent way. Thus, this chapter will deal with certain problems that have been thought to arise for the neo-Lockean, and the conception will be saved from certain potential inconsistencies. A study of the problems of amnesia shows how it is possible to use our conditions on survival to yield a surprising answer to a well-known puzzle. Since unity-claims can be made in the absence of experience memory, it turns out that there is no real ground for maintaining that memory is constitutive of personal identity. However, we can still maintain a development rather like Parfit's while acknowledging this. This result concerning amnesia will therefore be welcome to those who accept the Parfit line, for it means that the case of amnesia need not be assimilated to the quite different case of fission. In the rest of the chapter, I turn to the consideration of the problem which seems to me to provide the greatest obstacle to developing any useful theory of the person. Although we are used to thinking of persons in dual terms, as somehow being both mental entities and physical things, there is another dualism which also infects our attitude to persons. For ease of labelling, I call this a type-versus-token dualism. On the type theory of persons, a different story about bodies is given from that given by defenders of the token theory. In the end, the issue of the causal connections between selves is, I think, what is of real importance for survival. This issue cannot be resolved, however, as long as our notions about persons remain poised between the two kinds of theory identified. This means in particular that we cannot give any real support to Parfit's development of the notion of the person.

Amnesia and Other Problems


We have already noticed that the tenn 'continuity' is used in a somewhat strange way by some authors. In the Hollywood sense of the term, a series of episodes is continuous when each one goes on where a previous one left off, even if the episodes are separated in time. It is important to bear in mind that this is not the kind of continuity I intend when dealing with psychological continuity. Rather, let us forget-just for the moment-problems about episodic objects. Thus we can make sense of psychological continuity in the following way. Suppose that the mind is a structured bundle of thoughts, desires, hopes, wishes, intentions, fears, ambitions, memories, and so on. Now such a bundle could show either weak or strong continuity in the sense identified by Hirsch. If the bundle were weakly continuous, then this is compatible with its history's having 'jumps'. Recall Hirsch's example of the tree (Chapter 2). Although we may lop a very large branch from a tree, and the tree thus has a discontinuous history, the tree none the less shows a weak continuity in its history (provided its trunk and other branches do not have episodic existence). Likewise, some of my thoughts may vanish quite rapidly, and some new fears, hopes, and wishes be acquired in a rush. None the less, these rapid changes may be associated with underlying continuities that show only gradual change. So, for instance, habits of response, procedural skills, and sense of humour may all be relatively unaffected by even quite gross and sudden changes elsewhere in my psyche. Now this way of thinking about a continuous mental life is in sharp contrast to the sort of thing said by some authors. Think, for example of David Lewis's remarks: ... what I mostly want in wanting survival is that my mental life should flow on. My present experiences, thoughts, beliefs, desires and traits of character should have appropriate future successors ... Change should be gradual rather than sudden, and (at least in some respects) there should not be too much change overall ... Such change as there is should conform, for the most part, to lawful regularities concerning the succession of mental states ... 1

This looks very much like an appeal on behalf of strong continuity. But some fonn of weak continuity looks very much more plausible when it comes to describing the features of the mind over time. 1

Lewis [3]. The quoted passage appears on p. 17 ofRony [1].


Amnesia and Other Problems

Sleep, for example, is not accompanied by 'that consciousness' Locke mentions, which-as already remarked-is absent too when we day-dream, lose ourselves in absorbing tasks, and so on. During all such occasions, we lack a certain co-consciousness between that part of our mental life and its neighbouring stages. Notice again, that this talk of stages is not meant to suggest that our mental lives are discontinuous in time. It is important to recognize that Lockean self-consciousness may come and go, even if our mental lives are continuous in time. Even sleep, death's half-sister, does not interrupt the continuity of our mental lives, for we will register sounds, vibrations, and other disturbances while asleep and-of course--we also have dreams (whatever they are). We could say, then, that it is only from the point of view of my inner story that I appear to be strongly continuous. However, it is plausible, at least, to suggest that the history of my mind may involve only Hirsch's weak sort of continuity. Now Parfit recognizes that some of our mental episodes are connected with others (as a memory is to a past experience, or the formulation of an intention to its subsequently being acted on). These direct psychological connections bind or tie one mental episode, conceived as a psychological bundle, in an appropriate way to another such episode. What Parfit means by psychological continuity is nothing other than overlapping chains of such connectedness, that is, continuity is simply the ancestral of the relation of connectedness. Both connectedness and continuity of Parfit's sort will admit of degrees. This distinguishes Parfit not only from Locke, but also from a neo-Lockean like Wiggins, for both of whom co-consciousness seems to be an all-or-nothing matter. 2 On Parfit's account, we can quite sensibly inquire to what degree some earlier psychological bundle is connected to some later one. Let us now think of two stages in the psychological history of one person. In terms of the diachronic unity problem (as described in the last chapter), these stages can be thought of as bundles of psychological states connected, let us suppose, to some degree. If we imagine such connections being very dense, then we may find, for example, an abundance of memories in the one caused by 2 I have in mind Wiggins's discussion in Wiggins [7] and in the final chapter of Wiggins [9].

Amnesia and Other Problems


experiences in the other. In this case, what matters is-for Parfitthe richness of connectedness. As we have seen, it should not really worry us if there are, for a given earlier stage, two later stages that are equally richly connected. For although identity may be impossible to determine in such a case-or even be settled by stipulation-what matters is the psychological connectedness (with some appropriate cause) between the earlier stage and the later ones. Now the further apart any two stages are in time, the less rich the direct psychological connections between them are likely to be. Not only do memories fade, and intentions get forgotten, but even the relatively stable aspects of the personality show some change. For Parfit, we can have psychological 'continuity' between two such remote stages simply in virtue of the overlapping chains of connectedness among these and intermediate stages. In fact, he qualifies this. For he suggests that what we are likely to want, if we are to capture what is important in my survival, is that sequential episodes show strong connectedness. And this means that, let us say, half of the connectedness that normally holds between neighbouring psychological episodes, or that hold over every day in the life of the normal person, still hold. 3 Of course, strong connectedness is not transitive. But its ancestral is-and this is, as we have seen, simply 'continuity' in Parfit's sense. Parfit here is using the term 'continuity' in a way akin to the Hollywood sense already identified. This time, one later episode is not continuous with an earlier episode because it goes on just where that earlier one left off (for it may be very much later than the earlier one). Rather, it is continuous with the earlier one because it is linked to it by means of a chain of episodes, each of which went on (to a high enough degree) where its predecessor left off. Now it will be useful if we introduce a term that captures some of what Parfit is after by his talk of continuity, but which differs from it in two ways. First, we do not take it that the episodes in a person's psychological life are continuous in time. And, second, we allow that the kind of psychological connections we are interested in may amount to more than the sort which Parfit describes. Thus we count in personality 3 See Parfit [4], § 78. It is interesting to note that Parfit is not consistent in his use of the term 'continuity'. The present, and the following, section will show him operating with the term in distinct, ableit related, ways.


Amnesia and Other Problems

characteristics, skills, capacities, dispositions, modes of processing, and the rest, as well as memories, intentions, and the like. Let us continue to use Parfit's term 'connectedness' in preference to 'coconsciousness', only taking it now in the wide sense which embraces all the conscious and unconscious elements of a person's psychological states. And let us coin the new term 'coherence' to apply to the ancestral of the connectedness relation, noting that this means again that coherence can involve not just states of which we are consciously aware but the whole repertoire of our skills and capacities as well. Under this proposal, even if it turns out that persons are episodic in the way our hill of Chapter 3 was, we can still think of their persistence as coherent particulars. What I now want to show is that there is apparently a severe problem facing this kind of approach. For if Parfit wants to use connectedness and coherence in identifying what matters when we discuss personal survival, then he seems to fall foul of a dilemma. On the one side, it seems that disordered experiences satisfy his constraints on what matters even where the degree of disorder is enough to make us worry about attributing unity of person. And on the other, we can apparently think of cases where his constraints are not satisfied, yet we naturally want to talk of there being one unified person in their absence. Both sorts of problem are revealed by a study of amnesia.


Amnesia seems to give rise to two different problems. First of all, it gives us examples of coherence where there is a real puzzle over whether we have personal identity or unity. Consider, for example, the case of temporal lobe amnesics, as described in Iversen's wellknown study.' These unfortunate people apparently have reason4 In Whiny and Zangwill [1]. The effects of encephalitis may, in rare cases, produce even greater disorders than those described by Iversen. One case reponed recently involved a patient whose memory-span extended no further than a few minutes. Although able to apply concepts, play instruments, and even conduct choirs in complicated motets, his own view of his life was that he had just awoken from an anaesthetized state--a view he repeats reqularly. Along with this loss of even shon-term memory is a recurrent panic, which adds considerable distress to the patient's predicament-a distress he is able to experience afresh every few minutes. The referential practice of those involved with this patient, as well as the references

Amnesia and Other Problems


ably normal cognitive abilities-they can converse, apply concepts, describe situations, and so on. Additionally, they are able to acquire some motor skills, like the ability to carry out tracking tasks, and they will even improve these skills with practice. However, they do not have anything other than short-term memory abilities. Thus someone they have been introduced to one day would be greeted as a stranger the next. How, then, can they improve on motor skills with practice? Since they lack long-term memory, such patients themselves will be quite unaware that they are improving their skills, for on each new session of practice, they act as if the whole thing is quite new to them. However, those who are treating and working with such patients can make the observation that the motor skills are improving with practice. Moreover, such people show a high degree of psycho log ical connectedness from moment to moment; for their short-term memory capacities are not impaired. It is quite frightening, however, even to contemplate what life must be like for such impaired people. It is as if one's whole life is lived like a journey through scenery which vanishes forever as soon as one passes it. Others tell us that we have been at this point in the journey before, or that a particular piece of scenery is something with which we are already familiar. But to us, each experience can only be retained for a short time. In an analogy I have used elsewhere, think of a day in the life of such an amnesic as like a chalk line being drawn across a blackboard, with a duster an inch behind erasing all but the line's immediate past. In Parfit's terms, here is a clear case of psychological continuity, and so such an amnesic has a life which is coherent in our terms. Indeed, we can easily imagine that such an amnesic has a genuinely continuous psychological life (although my official position is one of agnosticism on this issue). Processing mechanisms can, as we have already noted, continue functioning despite sleep, day-dreaming, and even loss of memory. Yet there seems to be a real puzzle over how unified such a life can be. It might seem that the sameness of body displayed by a temporal lobe amnesic in some way compensates for their psychologically fragmented experience. More to the to him throughout this note show that we can operate with a notion of personal identity which is not tied to the subject's own retention of past experiences, nor to coherent attachment to certain intentions, loyalty to friends, or any of the other highlevel properties so important to the neo-Lockean.


Amnesia and Other Problems

point, however, we may argue that what is missing here is any high degree of certain forms of connectedness over time--something that Locke himself took seriously. Of course, the amnesic has had many experiences the day, or the week, before. But the inability to recall any of these to consciousness is what is striking. By contrast, the case with motor skills and cognitve abilities is not so very different from the norm. If we each have a distinctive cognitive style, a way of approaching problems, a characteristic style of describing things around us, then we can expect a temporal lobe amnesic to display just such a characteristic style over time as well. In the next section, I will in fact argue that, even without the aid of the body criterion, we can use these kinds of psychological coherence to support a notion of personal unity in such cases. A second and quite distinct problem that studies of amnesia provoke concerns the possibility of unity without coherence. Suppose that we consider a fictionalized case of amnesia where we imagine that a serious infection has caused someone to lose all memories of the illness and those extending back to some time before the illness as well. This is a case of what is known as retrograde amnesia. We can depict such a case diagrammatically, using the labels a, b, and c for extended psychological stages (Fig. 20). Let us suppose that common sense inclines us to allocate each of a, b, and c to just one person, Pamela. After t 2 , Pamela is unable to recall any of the events experienced in her b-phase, though she shows normal recall for events belonging to the a-phase. During her b-phase, however, and even during her illness itself (which, according to our story, occurred towards the end of that phase), she had normal recall of her a-phase. Thus there is connectedness between her a-phase and her b-phase, and between her a-phase and her c-phase. What is missing, and would be found in the normal case, is any connectedness between her c-phase and her b-phase. Now Parfit, and other theorists, assimilate the unfortunate Pamela's case to what is, in my view, a rather different one. 5 Taking seriously the view that Pamela's life after t2 is lived as if the b-phase has not occurred, they suggest that this sort of amnesia is rather like branching. In a branching case, of course, we would have to tell a rather different story. We would have to suppose that Pamela this 5

This is explicit in Parfit [4], ch. 13, and in Elliott [1].

Amnesia and Other Problems



tl - - - - -





I 1 ~-----




Fig. 20.

time underwent brain surgery of the sort already imagined in our earlier discussion of fission. Before surgery both halves of her brain were, let us imagine, brought into the same informational state. A donor body becomes available shortly afterwards, the hemispheres of her brain are separated, and the right hemisphere is transplanted into the new body. Her left hemisphere stays in Pamela's body. However, an infection proves disastrous, and this combination dies not long afterwards. However, the right hemisphere flourishes in its new body and this combination goes on to live for some considerable time. We can depict this kind of fission in the same way as we have done it before (Fig. 21). The diagram depicts only psychological relationships. This time, segment b represents the psychological history of the left hemisphere, while the more sucessful right hemisphere has its history shown by the c-phase on the diagram. Nqw, looking at this case, we can start to see why Parfit and others want to assimilate amnesia to branching. Think of the bphase asa branch-line, as Parfit puts it, while the c-phase continues Pamela's main-line. The branch-line is connected to the main-line to the extent that b is connected to a. But c is also connected to a,


Amnesia and Other Problems



Fig. 21.

while failing to be connected to b. The logical relations thus seems to be the same in this as in the previous case. Parfit gives a nice example of the assimilation of amnesia to the branch-line case. Notice, by the way, his explicit use of the term 'continuity' in its Hollywood sense. He writes: Certain actual sleeping pills cause retrograde amnesia. It can be true that, if I take such a pill, I shaH remain awake for an hour, but after my night's sleep I shall have no memories of the second half of this hour .... Suppose that I took such a pill nearly an hour ago. The person who wakes up in my bed tomorrow will not be psychologically continuous with me as I am now. He will be psychologically continuous with me as I was half an hour ago. I am now on a psychological branch-line, which will end soon when I fall asleep.6

If this is along the right lines, then we have a serious problem. For it does not seem plausible to maintain that the phases a, b, and c in our second case constitute the psychological history of a unified person. By contrast, there is real plausibility in the claim that we might make in the amnesia case that Pamela is-despite her amnesia-a unified person. Looked at in this light, we have unity without coherence. 6

Parfit [4), § 97.


Amnesia and Other Problems

The problem that emerges here is fascinating. If we can have unity without coherence, then there seems to be no good reason why we should not count the phases a, b, and c in the branching case as episodes in one unified life. But this looks absurd. For, in the branching case, while the b-phase exists, we surely have two people before us, namely the original-body person and the new-body person. And it is quite likely that they will not share all their experiences. Indeed, they may converse with each other, seeing each other from quite different points of view, explaining that each has different feelings from the other and so on. This assimilation of branching to amnesia threatens to make nonsense of that fact that although Pamela has, for a time, survived the fission surgery twice over, she has survived as different persons. By contrast, if we try to assimilate amnesia to branching, as suggested in the passage from Parfit, we seem to face an equal difficulty. For now, we have to count the b-phase in the amnesia case as being like the existence of a separate person within a larger life. But this means that what looks like the life of one person-including an episode with respect to which the person later has amnesia-is more properly regarded as the life of more than one person. This result strikes me as odd, to say the least. As a last, desperate attempt to salvage something from the comparison of the two cases, we might try to insist that there really is either connectedness or coherence, of the same sort, in both cases. Thus we might try to say that in both cases the b-phase is connected with the c-phase, via the a-phase. In the amnesia case, for example, there are memories in the b-phase of the a-phase. Likewise there are memories in the c-phase of the a-phase. So the b-phase is in some way indirectly connected with the c-phase F But then, we can claim this indirect sort of connection for the b-phase in the branching case as well. This, however, looks like a desperate recourse compared with what I shall argue is the most plausible way of dealing with our puzzle. For this involves simply separating the two cases. They are not, despite the similarities mentioned here, to be assimilated at all.

7 Thus Elliot, of precisely this sort of case: '(take) three slices of the one person, PI, andP3. PI is standardly, psychologically continuous with P3 and and P3 are not standardly, psychologically continuous. However, since n and P3 are both standardly, psychologically continuous with PI, we can regard them as indirectly psychologically continuous with each other.' (Elliot [I].)


n. n


Amnesia and Other Problems 9.3. WHY AMNESIA IS NOT A PROBLEM

If we recall our general conditions on survival, we can investigate whether they apply equally well to the fission case and to the amnesia one. Our question is to decide, then, whether the relation among the stages a, b, and c in each case is one of survival. Starting with the branching case, we note that in tenns of our conditions it seems reasonable to claim that a survives as b and that a survives as c. However, b does not survive as c. The reason for this is quite simple. Although c will-to begin with at least-show both structural and material similarities to b, these depend, causally, on a's prior possession of that structure and matter, but not in any way on b's prior possession of any such feature. Of course, to write in the way I have done here and earlier of structure and matter is perhaps to be guilty of over bold metaphor. Not everyone will agree that we can even begin to think properly about psychological phenomena as long as we persist in using analogies with ordinary physical things. One response to this is to point out that theory must be served. If we are to give survival theory a run for its money, then we must not cavil over this admittedly crude treatment of the psychological. And we may find, to our surprise, that we can go much further using this metaphor than we might at first have expected. However, in a more conciliatory vein, I might also point out that we need not think of the issue in tenns of structure and matter, but simply in tenns of deploying psychological analogues of the notions of structure and matter. We may have no idea what such analogues are; but this should not inhibit us from thinking about them. Nor, incidentally, is my usage here a great departure from common sense. If we do think-as neo-Lockeans tend t~f the human mind as like the software that perhaps runs on the body and the brain as its hardware, then we can reflect that the software notions like program, data, array, loop, and so forth are all easily subsumed under our labels. If we then return to our branching example, it is clear that the causal condition is the important one here. The brain transplantation operation, after all, used a procedure which involved copying the features of one of Pamela's hemispheres into the other. And it is this intitial copying that makes plausible the claim that a survives both as band c. We thus have a causal dependency of both band c on

Amnesia and Other Problems


a, but no more relation between band c than we expect to find in the products of a production process. Thus b does not survive as c. By contrast, we find that the amnesia case is very different indeed. In fact, there is not just one amnesia case, but rather a range of cases which we could imagine, We will select some samples from this range, and find that in all of them there is something which differentiates amnesia from fission. Suppose, first of all, that we take it that Pamela's c-stage is not connected at all with her b-stage, using the term 'connected' in Parfit's narrow sense. Thus we have connectedness of memory and intention from a to b and from a to c, but no connectedness from b to c. It looks at first sight here that we can have no grounds for a survival claim: just as in the fission case, b fails to survive as c. But we need to be careful here. Our causal condition requires that one stage survives as or in another stage provided the structure and matter of the later stage is causally dependent, in an appropriate way, on the structure and matter of the earlier one. But we can now see a dissimilarity between our two cases. For it is precisely the effect of the imagined infection that shows up in the amnesia about her b-stage typical ofPamela's c-stage. Her amnesia with respect to those parts of her life that occurred in stage b is causally dependent on things that happened to stage b. There is a complication here. For the infection that was imagined as bringing about Pamela's problems is not itself a psychological phenomenon. However, we can suppose that, since it has psychological effects, there will be some way of characterizing its impact on her psychological components, and their structures. Pamela's psychological bundle, at the time of the infection, would-let us suppose--consist, among others, of a number of complex monitoring, storing, and searching devices standing in various relations to each other and to devices linked to perceptual and central processing mechanisms. In functional terms, her memory devices would be responsible for consolidating, encoding, and thereafter retrieving information handled by the cognitive part of the bundle. s Suppose, then, that the infection somehow interfered with the 8 I am simply using---quite uncritically-the kind of model of memory assumed by the authors writing in Whitty and Zangwill [1]. Nothing of importance hangs on the question of whether this model is, ultimately, satisfactory. The imponant point to bear in mind is that any functional model of mind is likely to use notions to which the vocabulary of structure, matter, and cause is applicable.


Amnesia and Other Problems

mechanisms responsible for encoding information so that it could later be retrieved. In her c-phase, then, Pamela has no way of 'reading' the garbled material processed by her damaged encoding mechanisms. It is precisely this damage, though, that is causally responsible for her amnesia. The structure and matter of her cphase, then, is dependent on the structure and matter (including the malfunctionings) of her b-phase. Notice that in this case we are imagining less by way of dramatic structural and material changes than in the case, say, of the episodic hill. Quarrying at one hill episode, for example, could produce extremely impressive changes in later stages. Persons are living things, and since their psychological states, like their physiological states, change, then we might expect amnesia to involve rather more than just the usual sort of change. If perceptual and other processing skills remain unaffected, then amnesia will leave much of the perceptual and cognitive machinery of the subject unaffected. In Pamela's case, we have supposed that information is entered in the b phase, albeit in an unreadable form, so it seems safe to assume a pretty high degree of survival of psychological matter and structure from the b-phase to the c-phase. Here, then, we see a real difference between amnesia and fission, even when we imagine that there is no high-level connectedness between the stages in the amnesia case. But, as has already been suggested, the situation is likely to be rather different when we consider connectedness of a broad enough kind. For even though Pamela is not conscious during her c-phase of things done and experienced during her b-phase, due to the amnesia, there may well still be many features of her later psychological structures that are causally dependent on modifications to these during the b-phase. Take, for instance, her procedural skills. As was noted in the case of temporal lobe amnesics, skills will persist and improve even though the amnesic has no recollection of acquiring or practising them. Thus suppose that Pamela was keenly practising the piano just before her infection struck. In the c-phase, she can no longer recollect practising a certain new piece, but she can-none the less-play it much better than she would have had she not been pactising it during her b-phase. Thus, although hidden from her own 'inside' story, we external observers can note this kind of further dependence of psychological structure and matter between the two phases.

Amnesia and Other Problems


Obviously, thislast kind of information makes it rather clear that there is likely to be a real ditinction between fission and amnesia. It would be entirely surprising if, in the fission case, skills practised on the branch-line were to be transferred to the main-line. Of course, we could try to make some arrangment whereby this came to pass. For example, we could wire in tiny transmitters and receivers that would put the two separated hemispheres back in touch with each other. Or we could imagine that, rather like the children in The Midwich Cuckoos, anything that one of the separated hemispheres learns is also known to the other, perhaps due to some telepathic phenomenon that operates in such cases. As soon, however, as we make this move, we immediately change the fission story. It is now quite unclear what to say about the relation between the main-line and the branch-line persons. The very factors that made it seem so obvious that they were distinct persons are no longer present. In the light of my suggestions in this section, it follows that Parfit, Wiggins, and other neo-Lockeans need not worry about amnesia. Moreover, the application of our conditions to this case is so pleasantly natural that we seem to have found good reason for having a certain confidence in them. That they work so well on the psychological case, delivering a result that is congenial to good sense, suggests that they ought to be taken very seriously indeed. Even if we still maintain that our employment of the notions of structure and matter is metaphorical, the metaphor has proved useful. We might indeed want to claim that we now have good reasons for thinking that a chain of non-branching survival relations can underpin personal identity. Now, although it would not be wrong to argue thus, and perhap_s some readers will want to make this very use of the conditions, my own preference is to stay agnostic. If we recall Parfit's arguments about what matters, we would be wise to stick with the more modest conclusion that Pamela's stages show appropriate degrees of survival in the amnesia case. Thus a, b, and c merit being described as a series linked by the survival relation, the relation that matters. As well as complicating the amnesia story by imagining unconscious links between stages, we can also think of the impact of the body criterion. There seems little doubt that, from the outsider's viewpoint, Pamela's experiences, however disordered, can all be associated with one body (forget, at least for the time being, problems about the identity of bodies). It may be that, for all


Amnesia and Other Problems

we know, hamsters are incredibly amnesic. Each day may bring, from their interior viewpoint, radically new experiences uncontaminated by memory. But a child's pet hamster has, for all that, a clear identity-at least in the eyes of the child! The body condition seems to be what makes it clear in the case of fission that the mainline and the branch-line do not belong to one and the same person. For we have a great reluctance to allocate more than one body to a person at a time. Bodies do seem to loom pretty large in our everday notions of personal identity. However, the concept of a person that is yielded by development in favour of a body criterion is no more privileged, no more likely to be the right one, than any other. In the following section, I will make some suggestions about the weakness and strengths of the body approach. For the moment, let us observe how we have fared in dealing with the dilemma facing Parfit. As was noted in the first section, Parfit's constraints on connectedness and coherence seemed to permit highly disordered experience to be coherent. Extreme amnesia, where not even motor skills nor modes of response show preservation from day to day, would be an example of highly fragmented psychology that might none the less display connectedness over short times. But high connectedness over short periods, along with high loss of material and consequent structural changes over moderately short times, does not constitute the kind of coherent experience we assocate with a persisting person. So the first of the problems we identified in Parfit's position has not so far received a satisfactory solution. The fact that the disordered experiences are temporally continuous in no way forces us to count them as unified. We will ~erefore have to bear in mind, as an outstanding problem, the question of what to say in a case where this kind of disorder is present. What is more likely, of course, in a real amnesia case, is that we encounter conceptual skills that are able to develop, motor skills that can go on improving, but a poor degree of experience memory. Life for someone in this position is not so highly disordered as in the case just discussed. Rather, our extending of Parfit's notion of connectedness to embrace modes of response, skills and so forth, allows us to view this kind of life as showing connections among episodes-even episodes remote in time. The lack of conscious acknowledgement of these connections on the subject's part is perhaps not so important as the existence of the connections

Amnesia and Other Problems


themselves. Thus, from an outsider's viewpoint, at least, there are grounds for saying that we have enough psychological unity to talk about what matters being preserved. Failing fission and other oddities, we have something here close to a plausible account of personal identity of a kind. The other horn of the dilemma facing Parfit was that experience that failed to satisfy his constraints none the less could be described as unified-as constituting episodes in one life and being continuous in time. Our separation of amnes.ia from the case of branching has helped here. Within a unified life, as we have seen, it is perfectly possible for there to be little connectedness of memory, intention, and the like over even fairly short periods oftime. We are all of us amnesics to a great or lesser extent, for few of us can recall more than a minute portion of our past experience. Some of those who are amnesics in the clinical sense will fail to satisfy Parfit's connectedness constraints while still leading lives that are continuous and unified to a degree. By broadening the scope of connectedness and including what we might call 'low level' connections, we have given an apparently satisfactory solution to this part of the dilemma facing Parfit.


It was observed earlier that one development of our concept of a person involved associating persons straightforwardly with bodies. Although in this chapter we have been looking primarily at the prospects for defending Parfit's and Wiggins's forms of Locke's position, we have found once again that bodies are very much a part of our everyday notions about personal identity. Maybe, though, this reflection is not important. For, as we saw in the last chapter, Parfit has a very powerful case against taking the notion of personal identity as important at all. Since identity is liable to depend in some cases on essentially trivial facts, what matters is bound to be something different from identity. The liberation from concern with identity as the thing that matters need not, however, be a liberation from a concern with bodies. Although we can make sense of the neo-Lockean's concern with psychological relations between me now and other experiences and happenings in the future, we can also make sense of an equal


Amnesia and Other Problems

concern that what matters may be certain relations between my body now and some body in the future. Williams, as we have already seen, mentions the question of lO'Ve for a body. This he sees as a 'grotesquely misleading' way of describing one's love for someone, rather than as anything that might be philosophically mistaken. Bodily identity will, if Parfit is right, depend on trivial facts. But even if what matters when we are talking about the lives of persons is not identity, it may still be that what matters has some relation to bodies. Bodies have the useful property of being accessible to the outsider's view. Whatever it is really like to be me may be something that forever eludes elucidation or any sort of characterization in language. It may be an inexpressible thing-although none the less real enough for all that. But at least as far as my family, friends, and enemies are concerned, the existence of a living body with certain characteristics is pretty important to the persistence of me. As we have already noted, 'what it is like' to be Pamela, in either the amnesia or fission cases, may, when viewed from the inside, be pretty well the same in each case. But from our outside point of view, the association of memories, experiences, and behaviour with certain bodies is likely to be significant in our judgements either about identity or about what matters. We need to be clear that there really is something to be said here by those who oppose the neo-Lockean view-that they are not entirely silly to maintain that what matters might be possession of a certain kind of body, capable of doing certain things. One thing we should remember is that the very skills that persist through Pamela's amnesia might depend for their manifestation on her possession of certain bodily features. If she is a fine pianist, then she will have the muscle tone and other physiological characteristics that make good performances on the piano possible. If she plays the trumpet, she will have breathing control and diaphragm development that is different from the average. If these seem essentially trivial points, then we must remember that not everyone places equal stress upon the more psychological or intellectual aspects of personhood. It makes perfectly good sense to maintain that we like someone because they remind us of one of our friends by virtue of the way they speak, smile, walk, lisp, drink coffee, and so on. To the response that such liking is of an essentially inferior kind, as based on essentially inferior features of persons, we

Amnesia and Other Problems


have to ask what the standard is by which 'inferiority' is to be measured. Even if we were to concede that such matters are of less consequence than the more lofty-perhaps psychological-traits displayed by persons, we would at least have to admit that as a matter of fact people very often fall in love, set out on projects, face risks, and act in many other ways largely in response to their perception of such 'inferior' traits. It is not, in my view, implausible to maintain that we are tom two ways in our account of what matters because we do not know to what extent we value the physical and to what extent we value the nonphysical aspects of personhood. This is not, as might at first appear, because we are confused about values, although I am sure we are. Rather, it is because we have no clear conception of what a person is-and we are therefore at the mercy of somewhat inconsistent expectations. A body theorist can develop a consistent account, by pointing out that what matters to us is the existence of a body and the association with that body of certain behaviour. The same theorist can say that psychological characteristics matter only to the extent that they are responsible for the production of the behaviour which we recognize as typical of the person in question. Such a theorist, then, need not deny the existence of psychological or mental phenomena. Nor is the theory committed to viewing bodies as relatively unchanging, passive particulars. Rather, bodies matter to us by virtue of some of their individual features and the history that as tokens they carry; but the behaviour of the person we care about is also expressed via the body. To behave in the way persons do is to act intentionally, rationally, foolishly, cleverly, or whatever, and such action requires the possession of both mind and body. We can make some headway on this by reflecting again on the matter of daffodils. As we argued in Chapter 4.8, daffodils have the interesting feature of producing remarkably similar blooms each year. We can thus value a particular daffodil because it is a production process, or at least houses such a process. Each year, this process results in the production of blooms which we admire, savour, and enjoy. Likewise, our imagined body theorist might maintain that what we value in a person is their ability to produce certain kinds of behaviour, certain responses to situations, certain jokes, certain cheering words, and so on. What we call their psychological states are simply some of those properties in virtue of which they are able to make the responses they do to the situations


Amnesia and Other Problems

around them. Of course, being the sort of thing that can develop a repertoire of interesting behaviour typically involves having experience memory, ambitions, intentions, and all these other psychological states that the Lockean deems so important. For our imagined body theorist, however, these are only incidentally important-important as a means to producing those pieces of behaviour that are unique to and typical of persons, and for which they need to have bodies. Now the idea that persons are, or house, production processes, has far-reaching consequences. And it need not be regarded as the sole prerogative of the body theorist. There is, I think, a certain truth in the underlying insight. We do care about people's behaviour, and we do think of that behaviour as typifying them, as-in some way-individuating them. Thus, although we may mourn the death of an old friend, we are pleased to see some survival of that persons's integrity, humour, or courage in the behaviour of one of their children. Now imagine that experience memories were passed on from parent to child. How much would we count the parent as surviving in the child? The degree of survival would not be as much, I conjecture, as it would be if the child also reproduced the parent's way of talking, mode of gesticulation and other behavioural characteristics-like the ones just mentioned above. Recall that, in our earlier speculations on the survival of parents in children, we were able to build our case on precisely such behavioural commonalities. However, the body theorist should not build too much on the insight regarding behaviour. It is true that with a different face I could not smile one of my typical smiles or scowl one of my typical scowls. It is also true that with only one leg I could not cycle the way I currently do, and so on. But much of our behaviour is not body specific. Many different bodies would do for the production of gestures, the mimicking of accents, the waggling of eyebrows, or whatever. So although what matters, for the body theorist, might be the existence in future of a certain kind of body with a certain range of behaviour open to it, it is not clear that the theorist I have described here would be able to insist that what matters is the existence of some particular body. Indeed, in response to Williams's observations on the matter of loving a body, Parfit cites Quinton's view that obsession, or lust, for a body is usually directed at a body type, not a body token. In what

Amnesia and Other Problems


Parfit thinks to be the normal case of infatuation of this sort, the obsession is easily transferable to a replica (even an approximate one) of the original body. By contrast, love normally is-according to Parfit-concerned with the continuously changing mental life of the loved one. Furthermore, love develops and grows over time, involving a shared history, and thus memories as well as bodies. Moreover, Parfit, and other neo-Lockeans, are not maintaining any kind of commitment to disembodied existence. If what the body theorist is really getting at is that what matters for my persistence is the existence of some body, then this may well be acknowledged by all the software theorists. For without hardware to run it, software is peculiarly useless. In thinking about conceptual development, then, I will be taking at least this much on board from the body theorist. But this does not mean that the concepts of the person we will discuss are concepts of bodies.

9.5. SLEEP

We have seen so far how the neo-Lockean can deal with continuous and unified experience which is not linked over time by high-level connections. This is the experiential situation of the amnesic who is not so highly disordered as to lack coherence of a broad kind. In fact, our introduction of the broad notion of connectedness has made the position of Parfit rather stronger than it might first have appeared. Wiggins's position, however, is not so clear, for Wiggins uses the notion of co-consciousness in very much Locke's way. Thus Wiggins seems to be interested mainly in connections among episodes that are clearly conscious episodes. There is nothing, of course, to stop a defender of Wiggins dropping this requirement and operating instead with the broad conception of connectedness I have urged here. But this would be a substantial departure from Wiggins's expressed views. However, one more problem still remains for Parfit's position. As we noted, a high degree of connectedness from moment to moment, coupled with large losses of cognitive material and structure over slightly longer periods, would give us a set of experiences that seem to satisy even Parfit's modified unity constraints while raising doubts about whether we really do have unity. Suppose there are people so disordered that, although each day their recall of the


Amnesia and Other Problems

previous day's witnessings, learning sessions, and so on is close to nonnal, this ability declines rapidly over the space of just a few days. By Friday, their memory of what happened on Monday has almost entirely disappeared, although their recall of Thursday's happenings is perfectly nonnal. Likewise, although skills learned on Thursday can be improved on Friday, those acquired on Monday are apparently lost forever and need to be learned anew. It seems highly unlikely that people born with these defects would ever be able to acquire language or learn to operate with concepts at all. However, injury and disease bring about the most peculiar coqditions sometimes, and we can perhaps imagine that the patients in question have suffered some accident in adult life which has brought about the sorry state of affairs just described. 9 The nature of the disorder, then, is such that a victim of it can still communicate and deploy concepts to some extent, though since the accident no new skills have been gained nor knowledge acquired. Such a person no longer seems to have a coherent life, despite the connectedness fropl day to day. Put another way, it is hard to see why we should really talk of one person persisting since the accident, rather than of a series of short-lived persons who have existed since then. The notion that we are confronted in this case by a series of persons might be further supported if, along with the other symptoms so far described, the patient displayed radical changes in personality, behaviour towards others, moral views, and so forth. It is clear that, faced with this kind of dislocation of behaviour, our broadening of the notion of connectedness does not help resolve the problem. In fact, in tenns of our broad notion of connectedness, the patient clearly lives a disconnected existence. It seems to me that there are two things we might say about such a case. On the one hand, we might appeal to the continuity (or at least the persistence) of the person's body to support a claim of identity along with an admission oflack of unity. On the other, we might want to admit to an inability to settle the identity issue here at all. Whatever line we took, however, we would, if inclined to Partit's position, maintain that in terms of what matters, we are not now confronting a unified 9 Our a priori notions about how the components of our minds relate to each other tend to be flung into disarray when confronted by clinical accounts of actual psychological disasters. For some relevant case histories, see Sacks [1].

Amnesia and Other Problems


person. Yet again, there would be a gap between our decision on identity and our decision on what matters. We can see just why the Parfit-minded would take this line if we think of a fictional community. In this community, rather like ours, people generally go to sleep at night and wake up the following morning. However, their scientists have discovered that sleep for these people does genuinely interrupt their psychological lives. Their psychological lives are best thought of as sequences of episodes interrupted by sleep-in the way that some of us think that our conscious life of experience is interrupted by sleep. Now what would a coherent life for such people be like? The answer is that it would be very like the lives we lead. Each morning, just like us, they wake able to go on with projects laid aside the day, or the week, before, recalling their lunch-dates, fulfilling their teaching commitments, writing letters to their friends, and so on. In our case, we may think that-despite the episodic nature of consciousness-we have psychologically continuous lives. At least, I assumed this to be so earlier in the present chapter. But our imagined folk have no grounds for holding any such belief. So on what grounds can they maintain coherence of experience and of psychology in general? There seems to be four features which show their lives to be coherent in our sense. There are, first of all, our three features of survival. Each episode in the lives of such people survives in or as its neighbours. Their lives are thus built up out of a chain of episodes in just the way our flickering hill persisted as a sequence of causally connected hill episodes. Again, we can only take this position if we permit ourselves to talk of psychological structure and matter, but I am hopeful that what has been argued in this chapter so far makes such a mode of speaking not seem too outrageous, But a fourth feature, on which we have so far been silent, would also be imponant. Put negatively, this condition would insist that our imagined people do not behave like the disordered patient described· above. Or, in positive terms, psychological episodes relatively remote in time must show sufficient connectedness (as opposed to coherence) in order for them to be allocated to one persisting person. Lewis makes this sort of point about survival by referring to the fact that in a single, unified life there should not be too much change overall. However, Parfit is able to make much the same point-as we


Amnesia and Other Problems

might expect-without making implicit or explicit reference to identity. As he puts it: On my proposed way of talking, we use'!', and the other pronouns, to refer only to the parts of our lives to which, when speaking, we have the strongest psychological connections. When the connections have been markedly reduced-when there has been a significant change of character or style of life, or beliefs and ideals-we might say, 'It was not I who did that, but an earlier self'. We could then describe in what ways and to what degree we are related to this past self. 10

I would suggest that if we take 'connections' in the above passage in the broad sense identified already, then we obtain the most plausible version of the view. Of course, I am assuming here that our imagined people really are different from us. Our psychological lives very probably are continuous-although I am prepared to allow that this is not a question on which philosophy has the resources to pronounce. For us, only our conscious lives are episodic in this way. But it is clear that we can make the point about connectedness over time whether we ate dealing with a continuous or a discontinuous phenomenon. Continuity is not, after all, the issue. Rather, what we are facing is a problem about the degree of change which is compatible with the persistence of a single, unified set of Humean bundles, Parfitian self, or whatever. Like all other living things, human beings change as they grow and age. Some of our structures are relatively unaffected by these changes-for example, our gene complement. Others change in the lawful and regular way that is typical of humans-males lose their hair, all of us gain wrinkles. Our psychological matter and its structures likewise change. But there must be preservation of some structural invariances for there to be (and hence for us to recognize) persistence of the same human body.ll Likewise, there must be preservation of structural invarParfit [4], § 101. McCabe shows this point clearly by reference to simplified drawings of a face, and her article contains many references to the psychological literature: see McCabe [I]. The reader may quickly become convinced by spending a few moments with pen and paper. Draw some simple object like a house, a sailing dinghy, or a stick figure, using lines. Now redraw the figure using crosses, small circles, or any other variant, in place of some of the lines used in the original drawing, but being sure to keep the relative configuration of the parts. Compare the result with another drawing in which the original lines are preserved as far as dimensions are concerned, but are located quite differently with respect to one another. Whereas the drawing that 10 11

Amnesia and Other Problems


iances for there to be persistence of the same psychological bundle, or self. We noted already that there must be principles of change and of conservation for any kind of thing, these principles differing according to the sort of thing in question. What we have now discovered is that the same is true of psychological as well as of physical phenomena.


So far, this chapter has defended a position rather like Parfit's. The intention has been to show how a neo-Lockean position, with appropriate modifications, can give an account of personhood that is interesting, and develops some of the features to be found in our normal ways of thinking about persons. Adding complications, we noticed that Partit's own account was an account of what matters in some of our thoughts about persons and thus can be given in the absence of a decision on personal identity, or even where the decision on identity has favoured a candidate other than the one identified by the what matters criterion. Thus, even if we decide that our highly disordered patient of the last section is one person, this decision need not involve admitting that the patient therefore possesses the unity that matters, the unity of psychological history that we normally think a~companies personal identity. By contrast, if we decided that our imagined people with episodic psychological states really were new people each morning, this should not inhibit us from noting that these separate people display an important unity of psychological history. In this section and the following one, I want to conclude the treattnent of Parfit by showing that some problems remain for his view, despite its evident attractions. In the next chapter, we will consider whether a modification of Parfit's view will enable us to develop a concept of person that is genuinely neo-Lockean, and associates personal identity with what matters. Such a development would be greeted with some relief for those who find it a great strain

maintains structure will clearly be a representation of the same thing as the original, the drawing that maintains the lines but places them in a new configuration will either represent some different thing or be a mere pattern of no particular significance.


Amnesia and Other Problems

to bear in mind that what matters has to be kept separate from the identity question. . The current neo-Lockean position has treated selfhood as being rather like software, while denying that there is any way in which we can make sense of the existence of disembodied selves. What is important to the unity of self in normal situations is the survival of one psychological bundle in later ones, allowing for that degree of change that is appropriate to persons, or selves. In the unusual cases, we may not be able to talk of unity of self or person, but we can still recognize that different bundles-even when belonging to different persons-can show the connectedness and coherence that counts. We need to recall that the view is reductionist. Personal identity is not some deep or primitive fact to be argued about even after all reasonable information-collecting steps have been taken. It is for this reason, after all, that Parfit is able to disentangle the issue of identity from the question of what matters. If Parfit is right, then there are important consequences that follow from accepting his reductionist views. Take the example of death. Death is less terr~fying, perhaps, if we recognize it as no more than an extreme case of the loss of connectedness and coherence characteristic of normal life. On the positive side, I can be consoled by the thought that after my death there will be Humean bundles whose matter and structure is rather like the matter and siiucture of my bundle. If enough of my enthusiasms, obsessions, hobbies, and projects are still going on, then I have, to some extent, survived. Parfit writes extremely eloquently in favour of the reductionist view: When I believed that my existence was such a further fact, I seemed imprisoned in myself. My life seemed like a glass tunnel, through which I was moving faster every year, and at the end of which there was darkness. When I changed my view, the walls of my glass tunnel disappeared. There is still a difference between my life and the lives of other people. But the difference is less .... After my death, there will be no one living who will be me. I can now redescribe this fact. Though there will later be many experiences, none of these experiences will be connected to my present experiences by chains of such direct connections as those involved in experience-memory, or in the carrying out of an earlier intention .... My death will break the more direct relations between my present experiences

Amnesia and Other Problems


and future experiences, but it will not break various other relations .... Now that I have seen this, my death seems to me less bad. 12

Is Parfit right to be so optimistic? We an only assess this point, I think, if we give further consideration to the nature of the connections between my life and the lives of others. Let us think back to the case of Beatrice discussed in Chapter 3.5. Recall that she had an adventitious replica in some distant star system, a replica, that is, who was not obtained from Beatrice by any copying process. Now, when Beatrice died, she did not survive in Ol as her replica; for she played no causal role in the creation of that replica. We deferred further consideration of the case at that point, noting that we apparently had no good theory of types and tokens that could be given purely in terms of the theory of survival. It is time, though, to think again about the problem raised by the Beatrice case, for it can help to illuminate Parfit's larger claims. Thinking back to the distinction between production and copying processes, we will recall that both processes can yield tokens of the same type. However, in the case of copying processes, we were able to give a reduction of type-token talk as simply shorthand for what we could more lengthily state in terms of our survival conditions. But we had not at that point considered the feature about persons drawn to our attention by the body theorist of section 4 of the present chapter. We noted there that one reason for valuing persons was as items that housed certain production processes, whose products we value. One of my friends regularly cheers me, let us suppose, by virtue of a behavioural repertoire which is invariably optimistic, humorous, and caring. My friend can be regarded as carrying around a set of recipes for behaviour: it is the behaviour that we value, of course, and the recipes (usually what we call the person's 'nature', or personality) are valued just to the extent that they produce this very behaviour. If we stick, for a moment, with this admittedly crude picture, we can start to make some sense of something that Parfit may be after. For, in the passage just quoted, his suggestion seems to be that the lack of connectedness between me and some future persons is not the only thing to think about when considering my death. Maybe we should think also about the continued existence of recipes, or of 12

Parfit [4], p. 281.


Amnesia and Other Problems

behaviour produced according to recipes. The fact that there will in future be certain experiences and pieces of behaviour that might have been causally connected with my present experiences and behaviour may give us grounds for some consolation. In terms of our Beatrice example, maybe we should say that although Beatrice does not survive in her replica, her personality, or modes of behaviour, are at least reproduced in her replica. And although Beatrice's death is a bad thing, the existence of the replica makes it not quite as bad as we thought. If I am right in taking this interpretation, we have to consider just how to make the claim plausible. For, looked at another way, the fact that there is no causal connection of the right sort between Beatrice and her replica, or between me and some future person who has certain similarities to me, is precisely what makes the critical difference. We can get clearer on this by going back to the issue of q-remembering. Recall that both Parfit and Shoemaker hold that it is possible to q-remember the experiences of another. Memory thus displays the duality of token and type: I and another person can both share q-memories of some third person's experiences. We thus each have our own tokens of the same memory type. Now, in introducing the concept of q-memory, I suggested that we might supply some appropriate causal story to link my possession of the q-memory with the original experience. Thus we imagined machinery that copies information from one brain to another. Now does q-memory, as envisaged in this way, hold out prospects for the survival of the person having the original experiences? J ones, let us say, underwent a whole host of experiences many years ago which I am now able to q-remember. Q-memory is thus a boon for me: it enables me to extend my knowledge and experience in the same vicarious way that I can by going to the movies or by reading a book. But does it do anything for Jones? Perhaps our response is to say that it enables Jones to survive to the same extent that an author can survive through a book. The analogy, indeed, looks quite close. If it makes sense at all to say that authors survive through their books, then it makes just as much sense, apparently, to suggest that Jones survives through my presently q-remembering his or her experiences. We can contrast this kind of situation with the following one. A machine is devised for enhancing our range of experiences in a

Amnesia and Other Problems


world where real variety of experience is increasingly difficult to obtain. This machine 'writes' into the brain just the appropriate structural modifications that will make the subject appear to have memories of a whole range of things that the subject has never in fact experienced-climbing Mount Everest, piloting Concorde, giving birth, or any of a thousand other things. Using this machine, I find myself in the same state, qualitatively speaking, as I would have been in had I been given some of Jones's memories. That is, there would be no discernible difference from my point of view between having the ability to q-remember bits of Jones's past and having been broadened by the experience enhancer. But this is purely chance. J ones played no part in the programming of the experience-enhancing machine. Yet if I have used it to obtain these apparent memories, is there any sense in which Jones has survived? The response of someone attracted by Parfil's line here might well be to suggest that in this latter case, unlike the former, Jones has not survived. But, at the same time, the adventitious occurrence ofthings in me that are very like q-memories of Jones's past means that Jones's fate is not so bad as we might have thought. Just as in the Beatrice case, things are not so bleak as we might at first have been inclined to think. Beatrice did not survive in her replica, and Jones has not survived thanks to the experience enhancer. But there are connections between Beatrice and her replica and between Jones's experiences and my new apparent memories. These are not direct causal connections. But maybe the lack of causality here is perhaps not the most important thing. Let us call the apparent memories put into me by the experience-enhancer e-memories. Then Beatrice and her replica are two of a type, and your qmemories of Jones's past and my e-memories of parachuting over Senegal are tokens of the same type. And maybe this reflection might make us feel closer to others, and less concerned about death.


We can now attempt to bring the discussion of the last section and section 4 into clearer focus. There seem to be at least two dualisms at work in our account of what is important, or essential, to persons. First, there is the traditional duality of the mental and the physical, and second, the duality we might want to describe in terms of types


Amnesia and Other Problems

and tokens-although, as we shall see in a moment, this may not be the best way of describing it. If we think of a simple case involving an artefact, we can get some grasp of what the type-token duality involves. If! lose myoId, dogeared copy of Word and Object, my distress is to some extent mitigated by the fact that other copies of this book are readily available. Although these are not copies of my copy, they possess something that was distinctive oj, although not unique to, my copy. This is the ability to convey Quine's ideas on language, modality, and ontology. My copy's marginal notes and coffee stains were not doubt not only distinctive of it but also unique to it. A book type, then, can survive the destruction of some tokens of it provided other tokens continue to exist. Notice, however, that the term 'survive' was not used in the last sentence in the technical sense we have given it. Moreover, as we have already seen, our reference in some context to 'Quine's book' or to 'the book I am now holding' may be indeterminate between reference to a type or to a token. Suppose, then, that there is a duality not unlike this possible with respect to our talk about persons. To make a crude simplification, let us think about our reference to someone's typical smile-a piece of behaviour distinctive of that person in certain circumstances. Is this smile unique to that person, or simply distinctive of them without being unique? Let us associate with what we will call the type theory the claim that smiles may be distinctive of people without being unique to them. And with the token theory we will associate the contrary claim-that a person's smile is unique to them even although other persons may smile extremely similar smiles in similar circumstances. Let me emphasize again the point about terminology here. I am not saying that the distinction between the two theories really is best put in terms of types and tokens: the terminology is meant to be suggestive rather than accurate. 13 What goes for smiles may be thought of as applying to all the rest of a person's behavioural repertoire. So we can have a dual view

13 Basically, the type theory, as I here call it, is congenial to some of Parfit's and Quintan's views on bodies: that, for example, to become obsessed with someone's body is to become obsessed with a body type. The token view takes the person as very much grounded in their body as, in Descanes's phrase, 'constituting one whole' with their body.

Amnesia and Other Problems


about the whole complex of behaviour and response typical of--or characteristic of-someone. Notice that this is not to be confused with the duality of mental and physical. It is very likely the case that the kind of behaviour that is characteristic of persons as opposed to other things is dependent on their possession of various kinds of psychological structures. We will see later how one version of dualism attempts to associate all such structures with the possession of a brain. But this is not our prime concern here. The kind of dualism we are dealing with here is something completely different. The defender of the type theory would maintain that although persons have bodies, and thus concern with the health, looks, and other characteristics of these bodies is perfectly proper, it is only the association of this body with a particular person that gives us any grounds for caring about this rather than about some fairly similar body. The body is no more than the medium through which the person acts. Consider a person as having a set of production processes associated with them-recipes, as it were, for the production of behaviour, recipes which are responsible for their loves and hates, passions and hobbies, values and deeds. It is this behaviour which, in the end, is important. This body which I now have is, on this view, rather like the token copy of Word and Object, one item through which my production processes or recipes work to produce my characteristic behaviour. But if these recipes and pieces of behaviour could be associated with a quite different body, then I would be associated with that body. By contrast, the token theory denies all this. On that account, the type theory simply begs the question. What makes behaviour mine, after all, ifit is not just its association with this very body, my body? Notice that the token theory is not a token theory of behaviour. For someone attracted to this line can readily understand the claim that the same behaviour (type) can occur on different occasions. Rather, it is a token theory of the person. What makes two tokens of the same behaviour type both pieces of my behaviour is, on this account, their association with my body. ·The ownership of the recipes, then, is detennined by the identity of the body realizing the recipes which are made manifest through behaviour. There is no way nonnally, in which these recipes and their associated behaviour can survive in a different body. One qualification to be made here is that the brain theory of the person (to be discussed in Chapter 10.3) gives some account of how I could come to have a different body, while still


Amnesia and Other Problems

asociating my identity quite firmly with possession of one particular lot of physical stuff, namely my brain. It is interesting to note, then, that what I have called the 'type theory' is not committed to either of two famous doctrines. It is not committed to Cartesian or Lockean dualism about the mental, nor is it committed to the new version of this dualism, central-state materialism. Parfit, as a representative of the type theory, shows how it is possible to maintain that view while avoiding precisely these commitments. Any attempt to say we are dealing with a dualism about persons here is therefore using the term 'dualism' in a heavily modified sense. Now we might think that if the type theory does not involve some kind of dualism of the traditional sort, then it has little chance of impinging significantly on our views about what is important in life or what is significant about death. Getting to grips with this issue is of extreme importance to Parfit, for he maintains that his theory is capable of making a radical impact on our view of death and of what is of moral value. 14 If the type theory is correct, then there are possible worlds in which my present body (perhaps thanks to teleclone machines) is only the body of one of my tokens. Moreover, even in this world, connectedness between me now and what we normally think of as our past and future selves is a matter of degree. In some cases, the degree of connectedness is so low that we can think of me now and that self then as tokens, so to speak, of different types. Continuing along these lines, we can start to consider the issue of what degree of connectedness and coherence with some later self is necessary for the survival of my present self. In crude terms, my present self is no more than a complex of recipes associated with my responses, interests, desires, passions, loves, obsessions, and so forth. Now is what I want, in wanting the survival of my present self, that it should contribute in some causally appropriate way to the existence of some future such complex or bundle? If we are inclined to answer this question in the affirmative, we need to go on to consider just why the causal contribution of this present self is of any importance. The loss of one copy of Word and Object was not the loss of the book. And maybe the loss of this bundle of recipes is not the loss of 14

See, e.g., Parfit [4], §§ 102-8.

Amnesia and Other Problems


these recipes. Other persons-whether causally influenced by me or not-are custodians of recipe complexes too. Some of them will display behaviour that in some respects is very like mine. If we value behaviour, passions, pursuits, and responses of certain sorts, then the world will not be without the ones I manifest simply because the world is without me. Further, there are many ways that the world could in future be without me as I am now, only some of which involve my death. It may be that I change greatly through time. My current self may fail to survive to a very high degree in any future self because I have changed. We do not normally regard such changes in personality and behaviour as death, but maybe we need to revise our way of thinking here. Imagine one of my future selves, as we would normally think of it. It may have lost much, although certainly not all, of its connectedness with my present self. What relates it causally to this present self are the intervening overlapping chains of connectedness which I have called coherence and which Parfit calls continuity. Maybe there is nearly as much causal connection between me now and me then as there is between me now and some of my present students as they are then. In studying with me, they have acquired some of my passions and interests, and over time these have flourished. Causal connections, then, can hold between me and future selves that we would not normally allocate to me. This seems, then, to give us some means of thinking about survival as holding between me now and others then. We can thus come to wonder, finally, if the matter of causal connections can really be so important. We have already raised this issue in connection with the cases of Beatrice and Jones. If what we value are the recipes and the behaviour, then can we not, as type theorists, think of my relations to future persons with even more detachment than we have done so far? Think of it like this. I do not value the existence of future selves pursuing interests and caring about certain things because I have played a causal part in bringing about the existence of such selves. On the contrary, I may try to play some part in bringing about the existence of such future selves because I value the pursuits and interests themselves. But now, why should I care about the future existence of my students any more than I care about the future existence of other students? Is not what really matters, when I think about the future, the existence of people with the interests and concerns that matter most to me?


Amnesia and Other Problems

Such a way of thinking has certain virtues no doubt. It may give us the ability to view our own deaths with a certain detachment, and it may lead us to think more optimistically about the future. But I do not think that such a view really has anything to tell us about personal survival. For the very severing of the causal connection severs the link that is crucial in considering my future survival. If death seems less evil on this view, it is not because there are better prospects for my survival than we thought. In a sense, something distinctive of me can be thought of as surviving; but-as pointed out already-this would not be a case of survival in our technical sense at all. What we can get out of Parlit's view by way of claims about our own survival depends very much, I would argue, on the issue of the causal connection between my present self and certain future selves. But I think there is overwhelming evidence to suggest that the degree of survival between our own past and our own future selves is, for the nonnal person, far higher than the degree of survival that can hold between ourselves and others. This is hardly surprising, after all, if we consider that our recipes and behaviour are themselves at least in part influenced by things like our gene complement, a complement that is probably unique to each person. Our histories are also unique to us: even identical twins differ in their doings virtually every day. So to the extent that our recipes and behaviour evolve in response to our changing surroundings, our past experience, and the impact of others on us, we will each show much higher degrees of survival over our own stages than any of these stages are likely to show to stages of others. It would be satisfying, perhaps, to conclude this development of the Parfit arguments by recognizing some sort of survival relation as holding between some of our stages and at least some of the stages of others. But we have only been able to go along the developmental lines above by taking the type theory very seriously. We have so far found no reason for thinking that the type theory captures either all of, or whatever is most central to, our concept of the person. For we are forced back again to the problem facing us at the end of Chapter 8.2. We noted there that our everyday attitude to persons involved vacillation between the type and the token theories. On the one hand, we can see attractions in the claims of those who emphasize behaviour and recipes as being what matter. On such a view, as we then noted, it is hard for us to be sure we survive even to

Amnesia and Other Problems


a degree in our children. More accurate, for all we know, may be the view that our children simply preserve and manifest features which have been previously manifested in us. The real survivor on this view is culture, the society, the gene pool, the moral code, or whatever. If, by contrast with this view, we think of the other things we say about persons, we find that they are regarded as intensely individual items which can act on, and transform, their environment and themselves. To be a person on this account is to be more than simply a manifestation of various cultural, genetic, and other values. On this view, we can think of ourselves perhaps surviving to a degree in our children and in our books, but only to a most attenuated degree compared to the real thing-our survival through the changes and vicissitudes affecting these tokens here and now. As long as we are caught between these very different views, then we have no principled means of resolving the difficulties faced in this chapter. We can develop a Parfitian concept. But we cannot be sure that it is proper to apply it. The gloomy conclusion, then, is that we are as yet unable to give an entirely satisfactory account of what we value in life or of what is lost by death.


Concepts of the Person 10.1. BRANCHES AND PERSONS

In the last chapter we did our best to develop a concept of the person along Parfit's lines. It turned out that the enterprise was only moderately successful. However, we ended up giving an account of certain important relations rather than an account of personhood itself. The reason for this was simple. On Parfit's own account, what matters when we think about our future survival is not always identity. In this section, I want to give brief consideration to the prospects for developing a concept of the person based on Parfit's ideas. In order to do so, we need to consider two different possible worlds. First, we take a world where there is no fission, that is, where survival never takes a branching form. Secondly, we will look at a world which permits fission and thus gives rise to the problems associated with having more than one replica of some original around at a time. In the first case, we can imagine that telec10ning of persons takes place regularly, but always serially. That is, no original survives entering the teleport chamber, nor does the machine ever produce two copies of the original. Alternatively, like Nagel, we can imagine a world in which people, after their thirtieth birthdays, go for regular annual replication in order to maintain their youthfulness. The scanning replicator destroys the brain and body of the person who enters it, but produces subsequently a body which is exactly similar to the original, except that no ageing has taken place. l The resulting replica is also psychologically connected to the original in all the right ways. If we call such persons series persons, then we might wonder whether series persons would have all the important features of persons as we know them in this world. For Parfit, they have. Indeed, Parfit goes further than this and argues that this world, as we know it, actually contains series persons: it happens 1

See Parfit [5), § 98.

Concepts of the Person


that, the way things are here, the series persons and persons as we normally think about them are exactly the same. 2 It is hardly surprising that Parfit should take this line. We can go further, and argue that in the possible world described, the series persons are nothing other than persons, according to an easily developed concept. Using our conditions on survival, we can claim that each series J}erson consists of a number of stages, each of which survives to an extremely high degree as its successor. Of course, the scanning replicator will introduce slight discontinuities into the history ofthe persons in the imagined world. But this need cause us no more trouble than the discontinuities in the hill described in the fourth chapter. If a sequence of episodes linked by the survival relation constituted one hill in that case, then the sequences of episodes linked by the survival relation in our new case constitute persons. This development of the concept of a person is not only in keeping with Parfit's constraints on what matters; it is also highly natural given our earlier account of survival. It might seem a bit more puzzling to know what to say in the case where replication is able to take a branching form. Suppose we imagine, then, a world in which person fission occurs. In such a world, the teleclone devices allow existence of the original while still producing replicas. Perhaps only one replica can be produced from an original at a time. Thus, the history of certain individuals can be represented in a binary tree, as shown in Fig. 22. Here each labelled segment of the tree represents an extended person-stage, and our problem is to see if we can answer the question: How many persons are represented by the tree? At each fission point, only one of the successive stages is actually a copy of the preceding stage: the other stage simply continues that predecessor stage's history. We might think that this fact settles the matter. If c, for example, is just a spatio-temporal continuation of stage a, then we might say that a + c constitutes one person (by the usual body criterion), while a + b is no doubt a mereological individual, but hardly a person. This, though, would hardly do justice to the Parfit-like concept we 2 'No phoenix has ever existed. But there are many series-persons. These sentences are being typed by a series-person, me. They are also being typed by a person: old-me ... I, the series-person, staned to exist when Old-I the person staned to exist. And it is very probable that both will cease to exist at the same time. This is very probable because it is most unlikely that, within the lifetime of Old-I, teletransponation will become possible.' (Parfit [4], § 98.)


Concepts of the Person



Fig. 22.

are trying to develop. For stage b is, after all, related just as strongly to stage a by the relation that matters---connectedness and coherence-as is c. In terms of survival, a survives as b to no greater or less a degree than as c. Further fission, indeed, affects the replica stage, for stage c, as the diagram shows, was not very long-lived and has no successors. A natural move here, if we really are attracted by Parfit's account, is to take survival as supporting, indeed as constituting, identity. A branch, or maximal path, on the tree is any route starting from the original segment and finishing at the end of a stage to which no more stages are subjoined. Let each such branch represent the history of just one person. Then the tree above would represent the history of four persons, and each of these persons is no more than a maximal sequence of stages related by the survival relation. We might, to take account of Lewis's point about there not being too much overall change, add to the characterization. The addition would reflect the importance of connectedness over reasonable lengths of time. As well as stage collection a + b + e + g being a maximal Srelated sequence of stages, for example, there would also have to be sufficient connectedness between a and g in order for the collection to constitute one person. This proposal may seem a bit drastic when it comes to the matter of knowing how to count, or speak about, persons. We usually think that if we count the number of person-stages present in a room at a time, then we have counted the number of persons in the room. But

Concepts of the Person


under the present proposal, this would not be true. If we imagine a time when stages b and c are alone in a room together, we would be making a mistake to claim that there were only two people in the room. For although stage c is unique to just one person (namely a + c), stage b is shared by three branches of our tree. However, in a world in which fission was commonplace, we might well have to get used to speaking in a slightly different way about persons. David Lewis makes some interesting suggestions about how we should do this using the notion of tensed identity (which is not any form of real identity). 3 If we assume that we could find some way of being comfortable about describing this world of fissioning persons, then we might also consider a new solution to our old case of Theseus' ship. In our earlier discussion of this case and of the 'only x andy' principle, we were happy to allow that the world might contain the mereological individual consisting of the original ship-stages together with the stages produced by the plank-hoarder. The present suggestion about persons, however, goes further by suggesting that in a world in which fission is commonplace we might even come to regard that individual as itself a ship. This would mean allowing that two different ships could have shared a common set of stages-not really very much more dramatic a suggestion than the one made about mereological individuals sharing stages. 4 However, even making these moves may seem not to give us, in the end, a concept of personal (or ship) identity that captures what matters. For although we are allowing a maximal aggregate of appropriately connected stages to be one person, it does not follow from this that psychological connectedness will on its own constitute identity. Even though we now have a means for determining that our diagram above shows the history of four separate persons, it might none the less be argued that the survival relation can hold between different persons. Some abbreviations will help in the discussion of this issue: Let us designate the relations that constitute personal survival 'the survival relation'. The survival relation is thus a relation between See Lewis [3]. Of course, we might be happier about wearing the complications of counting by tensed identity for individuals in the technical sense, than about doing so for everyday things. Individuals, unlike everyday particulars, are not, after all, generally of sorts or kinds. 3



Concepts of the Person

stages where one survives-according to our conditions-in or as the other to a suitably high degree. Now, as just proposed, a maximal set of stages connected by the survival relation and with not too much change overall will constitute the history of one person. Let us call, by contrast, that relation holding between stages only when they are stages of one same thing the 'identity relation'. In the paper cited in note 3 of the present chapter, Lewis argued that the survival relation is the identity relation. But that conclusion does not follow from anything we have said so far. Indeed, Parfit has argued in response to Lewis that even if we allocate personal identity as suggested in this section, we are still forced to recognize that the survival relation can hold between different people. 5 So if the survival relation is the one that matters, then we will find that identity and what matters are still separate. It is important to get to the heart of this disagreement, for although we have encountered several cases where identity and survival diverge, it is not clear that every case of fission needs to be added to the list. To get clear on Parfit's argument, let us think about the person in Fig. 22 constituted by stages a, b, and d-PI for short-and the person consisting of a + c-P2 for short. Now PI and P2 are distinct by our identity criteria, although they both share the common stage, a. If the survival relation were just the identity relation, Parfit claims, then we would get the following absurd result. P2 , during the shared a stage stands in the survival relation to stages c and b. But if the survival relation were the identity relation, a, b, and c would then have to be stages of some same person. But they are not. For whereas a and c are stages of P z' a and b are stages of the different person PI . So--according to Parfit-it does not follow from the fact that a person is a maximal aggregate of survival-related stages that identity coincides with survival and thus with what is important. However, in his later defence of the coincidence of the two relations, Lewis denies, in effect, that survival is ever a relation between distinct persons. With certain reservations, mainly about the nature of the branching envisaged, I have to side with Lewis here. On my 5 This was originally claimed in his response to Lewis [3], published in Rorty [1] as 'Lewis, Perry and What Matters'. In a new postscript to 'Survival and Identity' (found in Lewis [4]), Lewis has subsequently defended the claim that survival is constirutive of identiry. It is this claim which I support in the present sectionalthough only subject to certain rather large provisos.

Concepts of the Person


official view, survival is a relation between stages, and as long as we stick to this official view it is hard to see how to formulate the very claim that Parfit wants. His claim is that P2 survives both as P2 and as PI' The grounds for this claim are that stage a is common to both persons. Now suppose that during that very stage P2 (and hence also PI) had a thought along the following lines: 'I want to go on living.' There is an immediate difficulty about attributing this thought to anyone person (stage a is, after all, shared by four persons). Any argument to the effect that one of them had the thought shows equally well that another of them had it too. So in a world where fission is commonplace, we not only have to count by tensed identity, in the style of Lewis, but we also have to engage in radical reinterpretation of desires and other shareable states. Suppose, however, we agree that stages, while not themselves persons, can have desires. Then we can construe the thought above as indicating a desire on the part of stage a to be such that some person of which it is a stage should persist. We can try to put the matter, equivalently, in terms of persons. What PI' P2 , and the other stages share is the desire that at least one of them should persist. If we accept these proposals by Lewis, then we do have a way of preserving the claim that the survival relation is the identity relation, and hence saving the insight that survival is constitutive of identity. Of course, Lewis does not give my analysis of survival. So I have been foisting my account of survival on to him in giving the above paraphrase of his debate with Parfit. Even so, it cannot be argued that in a world where fission occurs survival would reasonably be constitutive of identity. In the case just studied, we have made a big assumption, namely that the branching is equal, in the sense that each immediate successor of a given stage has as good a claim as any other to be the survivor of the stage above it. But this notion of equal branching would only be applied to the case described for Fig. 22 by someone who accepted psychological connectedness and coherence as what matters. For someone who was impressed by the persistence of the body through stages a and c, the branching would appear to be unequal, despite the close psychological relations between a and b. It is precisely our inability to pronounce on what matters for personal identity that drives a wedge between identity and survival. Likewise, faced with the intractable problem of


Concepts of the Person

artefact identity, we were still able, in virtue of our three conditions, to say sensible things about survival (for example, that one artefact had survived as a different one). So the endorsement of Lewis's position requires major provisos. Provided equal branching occurs in a world of fissioning particulars, there is at least a theoretical possibility that survival in my sense is constitutive of identity. In a world without fission, but containing replication, the situation is, as we have seen, rather different. In such a world, our Parfit-inspired concept of a person is a natural development. That we are engaged in developing a concept in the case is shown by the fact that authors like Nagel would prefer to use a different terminology-namely that of series person-to describe just the same state of affairs. To the extent, then, that our world has in it analogues of replication and fission, we can talk about developing a concept of person, and an associated concept of survival, for its description.


It has already been suggested that we can develop concepts of the person at various levels. The kind of concept developed along Parfit's lines has been of an intermediate level. Although what counted as important in developing such a concept were psychological relations, these were of a very broad sort---on our modification, at least, if not in Parfit's original account. The relata of the survival relation, in other words, are, on such an account, psychological states, episodes, and data conceived as having structure and being capable of entering into causal relations with other such states and episodes. In this way, our conceptual development so far has avoided working up any really high-level concept of the person, though it seems clear that both Locke and Wiggins are interested in just such a concept. For them, with the emphasis on co-consciousness, the relata of the survival relation are conscious, or even self-conscious, episodes, which are in communication with each other thanks to memory, intention, and other high-level psychological operations. Such a high-level conception, of course, ensures that persons are episodic for, as we already noted, consciousness and self-consciousness come to us only periodically, with gaps introduced due to sleep,

Concepts of the Person


absorption in fascinating tasks, day-dreaming, and so on. However invisible these gaps are to our inner viewpoint, it is not hard for us to realize that they exist. We are aware of having 'lost ourselves', as we put it, in some interesting activity, or of losing our bearings when we drift off to sleep in the middle of a tedious meeting. On this high-level conception, one conscious episode survives as another, when connected to it by memory, intention, character, and so on, to a high enough degree. Notice immediately that such a refined conception of the person has two consequences. First, that it does not matter what changes there may have been in the structure and matter of the body that supports these conscious states. What is important is the functional relationship of items constituting the conscious states. Indeed, in an extreme case, we could even imagine reading off all the information concerning a conscious episode, loading this into a massive computer and then getting the computer to produce further episodes linked to this one. Here, we would have changed the entire physical means of producing such mental episodes. But for the functionalist operating with such high-level concepts of the person, such manipulations would be of consequence only if they destroyed important functional relationships. The notion that my present stage might survive (thanks to some elaborate recording and computing) as a stage of consciousness of some machine is not unique to the Lockean. On the intermediate account, developed from Parfit's, we could make sense of the same phenomenon, as long as the computer program provided survival for enough of my subconscious psychological states and traits as well. Indeed, any conception that takes psychological features as being what matter for identity or survival, and which avoids mystical appeals to the existence of a soul or spiritual essence, will be able to make sense of my surviving in or as a computer program. For the psychological account of the person is the software account. However, it is in considering this software account of the person that we see the real implausibility and weakness in the Lockean account. On the very high-level conception, not only is it possible to think of psychological states surviving physical rearrangements of the body, or transfer of the psyche to magnetic tape, the conception is also indifferent to changes in the individual's psychological configurations as well. Think of our characters, personalities, modes of response, memories, intentions, and so on all as the


Concepts of the Person

products of production processes. These processes manifest their existence by our characteristic forms of speech, behaviour, and action. Now, for all that I or you know, the psychological matter and structure that support such processes may be able to undergo massive changes, while still supporting processes which produce just the same product! Just as the same conscious episodes could be supported by differently configured bodies, so too they might be supported by differently configured minds. A certain kind of operationalism about mental states-emphasizing the way Dennett does the critical links between mind and behaviour-would no doubt find all this quite congenial. So long as the nature of the conscious episodes experienced seem unchanged to me, so long as the behavioural repertoire of my successive stages changes only gradually and along clearly intelligible lines, then my identity need not be bound up with any sameness either of bodily structures or of psychological structures either. What is wrong with this approach is, quite simply, that it flies in the face of our everyday experience and our whole method of approaching the explanation of behaviour. We do not theorize about persons on the basis proposed. On the contrary, we try to find diagnoses of personality disorders by looking for disease or injury affecting specific portions of the body; we recognize that after drinking too much alcohol, someone's reflexes will be impaired, and his or her judgement affected. These are not trivial aberrations about explanation on our part. Rather, they reflect a deep-seated conviction that our mental states and behavioural responses depend on quite specific production processes, any change in which will have significant consequences for our mental life and behaviour. This is not, of course, to deny that different creatures from us might enjoy similar mental lives and behavioural repertoires supported by quite different processes. But the evidence all around us is that the possession of the mammalian nervous system, the operation of systems controlled by the glands, and the development of the brain to a certain degree is precisely what gives the higher animals with which we are familiar their sophistication of response, and whatever inner mental life they may have. The Lockean conception's weakness, then, is that it stands in a certain tension with our other current views about the nature of explanation in the life sciences. This is not, obviously enough, a knock-down argument against the position. Nor would we expect

Concepts of the Person


any respectable philosophical position to be open to knock-down objections in any case. But these considerations do suggest that lower-level conceptions of the person, although they may do less justice to what seems important on the 'inside' view of personal identity, are likely to be more in keeping with our general style of approach to explaining the world we live in. So let us now descend lower than the intermediate conception of personhood in order to explore the prospects for developing a low-level conception of the person. Bearing in mind Williams's remark that our response to other persons (particularly our emotional response) reflects a deeply body-based situation, we might expect to find that some low-level concept of the person would reflect this situation. Surprisingly, however, it turns out that the most plausible of these low-level concepts still fails to do justice to this feature. The simplest low-level proposal about personal identity makes use of our old intuitive idea of continuity. Provided we have continuity of enough of a living human body, we have sameness of person. Thus persons are nothing other than living human bodies. This conception has at least administrative convenience on its side. In the psychiatric or neurological ward, where there are patients with dreadfully disordered lives, we can still apply such a criterion of personhood. We can thus count as one person a patient who suffers from multiple-personality syndrome, or one of our unfortunate amnesics with no long-term memory. Of course, there will be problems aplenty in deciding just how much of the body has to continue, or how much damage to the body can be tolerated, for identity to be sustained. But, to some extent, these problems are no worse for human bodies than they are for other living things. And our survival conditions can be brought into play here, for they do not simply apply to the case of episodic objects. From day to day, we should be able to find a high enough degree of survival of one bodystage in its successor in order for there to be continuing sameness of living body. This administrative concept of the person does do some justice to our body-based situation. To the extent that our concern for others is a concern for them as individual tokens, we will be concerned for the fate of their bodies (however odd this way of putting the point sounds). But concern for others as tokens does scant justice to that other dimension of concern we have for persons, namely our concern for those aspects of them which could be supported by


Concepts of the Person

other bodies. Nor is it clear that, in the case of multiple personality, amnesia, or other unusual conditions, we are right to follow the body condition. Telecloning fantasies only raise in an extreme form some of the problems associated with these real-life cases. However, there is a slightly different account of personal identity which we can offer here, which keeps some of the administrative virtues of the simple body theory while maintaining a conciliatory attitude to the claims on behalf of the importance of psychological relations. This is the version of the body theory that allows the brain, or the brain and certain other parts of the nervous and regulatory systems of the body, to be the components of greatest significance in establishing identity. We noted that daffodils, just like other living things, have components that are extremely significant for their identity, and others that are not. The loss of a leaf or of a blossom is not destructive of the identity of a daffodil. Likewise, the loss of a finger, or even an arm, however much of a handicap it might prove, is not destructive of the identity of a person. Just as the bulb seemed to have a crucial role for the daffodil's survival, so the brain, and other central controlling systems in the body, seem to have crucial roles. Moreover, the identification of the brain as the crucial component in survival has a further virtue. Our best scientific theories of the nature of personality, memory, perception, conceptuse, and so on associate all these faculties with the brain. Without a brain, I would not have those characteristics so important to my being a person. Moreover, it seems plausible to maintain, at least if we incline towards materialism, that as we change and mature, so there are related changes in the matter and structures of our brains. Notice that this focus on the brain as the important thing does not go all that far towards capturing what is important as far as the Lockean is concerned. If the identity of the person is associated with the persistence of his or her brain, then the best we can do towards characterizing fantasies about the migration of minds is to engage in fantasies about the transplantation of brains. This isfrom the point of view of those attracted by high-level accountsonly a bit better than the crude body account. It does, however, give us a way of constructing something like Locke's puzzle case. For we can, if we allow ourselves to speculate about brain transplantation, make some sense of the same person acquiring a new body. If materialism is true, then the focus on brains as the crucial thing, as

Concepts of the Person


being the things that matter, neatly welds together physical and psychological accounts of personhood. Such a materialist account gives us a dualism that is modified with a vengeance. The duality of mental and physical is reduced simply to the duality of different physical things. Notice, though, that the duality of brain and body should not be confused with the duality-noted in the last chapter-between behaviour and body, where behaviour is a repeatable, type phenomenon, and the body is a unique biological token. On the latter duality, brains are as much tokens as any other physical particular and as much as any other pans of the body. Brains are not only thought of as supporting very high-level conscious episodes. They also, so we think, support all the mundane processing, co-ordination, and problem-solving that any intelligent creature engages in, whether aware of this happening or not. So there looks to be a happy rapprochement possible between this version of the body concept and the broad psychological account deriving from our version of Parfit. It may well be-and I am not here trying to deny this for a moment-that in the end, our most useful concept of a person is to be developed along precisely these kinds of lines. Such a development would explain why it is, as we have seen in the last section, that the important survival relation can hold between two different people. For if brain hemispheres can be brought into similar informational states, then we could transplant one hemisphere into a new body and that body would then produce behaviour distinctive of-but no longer unique to-the original person. In fact, this procedure for bringing about person fission has already been suggested earlier. In using it, I was tacitly appealing to a view rather like the one now being canvassed, in the expectation that most readers would be prepared to consider as a case of fission one where the obvious carrier of identity (the brain) itself is split, and in so doing carries what matters (psychological connectedness) to a new body. I have previously identified two kinds of concern we have for others, and for ourselves. One is the concern with phenomena of a type (behaviour, moods, styles of response, feelings, and so on), while the other is concern with token phenomena (our bias towards specific bodies). Now it would be convenient if the notion of a person as individuated by the brain were to reconcile these two perspectives. That is, it would have helped our conceptual


Concepts of the Person

development a great deal if our concern with persons as tokens could be properly described as a concern with their brains. But this, very clearly, is not the case. The trouble with our dual concern is that to talk, for example, of our love for someone, our urge to be with them, our desire to touch and hold them, and so on is not readily translated into talk of love for a brain, or a desire to touch and hold a brain. On the contrary, our interest in bodies is far more closely tied to external features of the body. If we had to choose one bodily feature, it is probably true to say that most of us would find the face by far the most important, and significant part, of another person's body. It is not only grotesque, but also definitely misleading, then, to imagine the brain as being in any way the focus of our interest in others as tokens. Since it is relatively easy to imagine brain transplantation that moves the brain of someone we love into a body of a different sex, age, shape, and so forth, it is clear that the brain-based concept of personhood, like the higher-level concepts, captures some of the things we believe and think about persons without capturing all of this.6 By contrast, the low-level body account does justice to our response to persons as tokens, admittedly, but only at the cost of ignoring our interest in persons as producers of types of behaviour and response. But notice that the duality of type and token persists even when we give a materialist account of the mind. It is the existence of behaviour, of action (including language), of mood, and so on which gives rise to our difficulty . We value persons not only for what they are as bodies but for what they are as producers of behaviour. None of the concepts so far developed do justice to this duality. Yet it is the very existence of such a duality that leads us into some of the hardest puzzles about personal identity (to the extent that these are different from puzzles about artefact and animal identity). Our everyday talk about persons can now be viewed in the light of what has just been argued. We normally operate with a concept of personhood that is unexamined. It is used variously to apply to very different kinds of situations-situations in which we talk about our feelings for others, situations in which we discuss our beliefs about the differences between humans and animals, situations in which I

6 See, e.g., Puccetti [1] and the response in Brennan [1]. Needless to say, my views on persons have changed greatly since writing Brennan [1].

Concepts of the Person


we ascribe responsibility for actions, and so on. One part of the web of rules, associations, and inferences associated with this concept is that persons are rational beings, endowed with high-level abilities, like memory, the ability to intend, to reason, to be aware of some of their own inner goings-on, and the ability to speak in highly sophisticated ways. Another part of the same web associates value with the possession of person status: persons are not only different from other animals, but they have rights and duties which the other animals do not. Persons are also proper objects of caring, love, and regard-but this time, we admit that other creatures can also be objects of these things. The difference is more one of degree than kind. What may seem to have emerged from our argument so far is that no one thing can possibly satisfy all these desiderata. So our everyday concept of the person is perhaps confused, or vague, or indeterminate. In such a situation, can we wonder that no development of the concept of the person meets all the constraints we want to put on it? For these constraints themselves apparently derive from everyday confusion-not to put too fine a point on it. Is there then not something to be said for pressing ahead with what seems the most sensible form of conceptual development? Maybe, for example, we should give up our view that the object of my love-as object of love-is a person, reserving the concept of person for items that are rational and possessors of a special sort of value. But the people I love are rational, and possessors of a special sort of value. I also love things other than people. I love certain bits of nature, and certain books. But this does not confuse me. I do not think the natural things I love are rational items (I am sure trees are not rational), nor do I confuse the value of books with the value of my friends and loved ones. We cannot, in other words, press ahead with the most sensible form of conceptual development simply because we have as yet no grasp on the question of which development of the concept of a person is more sensible than any other. Such a conclusion is not meant to undermine the efforts of the Lockeans, or Parfit, or even the body theorists. It is important, after all, that theorizing goes on. For in a situation of confusion, we are naturally driven to seek clarification. My conclusion, then, should be read as an interim one. In the present situation, we have several developments of the concept of a person available. These are in competition with one another. But none is yet worthy of commen-


Concepts of the Person

dation over its rivals. Much more theorizing and exploration in conceptual development may be necessary before we can get any further in the area.


Given the attractions of the brain theory of personal identity, it is not surprising that some theorists have spent some time arguing for its plausibility. As with all conceptual development, they are on firm ground here. For the problem of development is not whether we can work up plausible concepts, but rather whether these concepts are more useful than their rivals. However, in the spirit of contributing to the theorizing that is essential to this area. I will look briefly in this section at one aspect of a brain theory. This is Nagel's unpublished account, discussed by Parfit in an appendix to his book.' Nagel defends a reductionist account of personal identity. Where he differs from Parfit is by holding that I am essentially my brain, and that the identity of my brain is what matters. It follows, on Nagel's account, that destruction of my brain is the destruction of what matters. Thus I am unable to survive a process which involves destroying my brain. Now, as must be clear, Parfit is bound to disagree with this. What I want to show in the present section is simply that we can defend the view that I can survive the destruction of my brain, while staying closer to Nagel's view than to Parfit's. The reason is that Nagel has not noticed the importance of our three conditions on survival. Indeed, the case that Parfit introduces in order to undermine Nagel's account is one which need not be described as the destruction of the brain at all. Parfit is concerned to show that the identity of the brain need not be what matters when it comes to the survival question. Instead of directly discussing Nagel's account of telecloning, he takes two cases which differ by less than the difference there is between destruction of the brain (and production of a replica by telecloning) on the one hand and ordinary continuation of the brain on the other. Parfit imagines that all his brain cells have a defect which, in time, would prove fatal, but which can be rectified by surgery which 7

See Appendix D of Parfit (4).

Concepts of the Person


replaces each defective cell by replicas which lack the defect. He considers the following cases: In Case One the surgeon performs a hundred operations. In each of these he removes a hundredth part of my brain, and inserts a replica of this part. In Case Two, the surgeon follows a different procedure. He first removes all the parts of my brain and then inserts all of their replicas.

He then describes the difference between these two cases in the following terms: In Case One, each of the new parts of my brain is for a time joined to the rest of my brain. This enables each new part to become part of my brain .... In Case Two, things are different. There are not times when each new part is joined to the rest of my brain. Because of this, the new parts do not count as parts of my brain. My brain ceases to exist.

Unfortunately, things are rather more complicated here than Parfit suggests. He wants to go on to argue that the cases pose Nagel a problem. Since the first case does not involve destruction of my brain, while the second does, Nagel is forced into defending the view that I survive in one case but not in the other. And this cannot be right: Can my fate depend on the difference :.n the ordering of removals and insertions? Can it be so important, for my survival, whether the new parts are, for a time, joined to the old parts? .. He [Nagel] suggests both (1) that identity is what matters, and (2) that he is his brain .... When I consider Cases One and Two, I find it impossible to believe both (1) and (2).8

But although we might at first be inclined to agree with Parfit here, we will come to a different verdict if we go back to think about simple cases involving replication. As we saw earlier, there are great problems about artefact identity. Indeed, our notions about artefacts make it very difficult to come up with a decent theory of identity for them. At least some of the time, we allow that quite massive replacements of an artefact's parts may not interfere with its identity. Recall the case of the car enthusiast described in Chapter 6.3. All sorts of part replacements were put down to the processes of 'modifying' or 'hotting up' the original car. Maybe Parfit has this sort of case in mind when he 8 These, and the previous quotations from Parfit, all come from Appendix D to his book.


Concepts of the Person

writes that the gradual replacement of parts results in the new parts becoming part of the original object. Hirsch suggests, rather optimistically in my view, that so long as we replace no more than a third of an artefact's parts at one time., then we do not destroy the artefact. 9 So I can replace a third of my car's parts on Monday, another third of them on Tuesday and the final third on Wednesday without having destroyed my car. Now what we need to distinguish in this sort of case is the difference between replacing old parts with new parts and replacing parts with replicas copied from the old parts. We saw before that there were a number of different things covered by our use of the term 'replica'. Thus we distinguished replicas produced by copying both from adventitious replicas (produced by chance) and from replicas produced by production processes. So let us make clear here just what the difference in this context amounts to. If I am to replicate by copying the shock absorber that has just been removed from my car, I need to engage in a manufacturing process that uses the original shock absorber as its prototype. My process, then, involves a copying process as part of it. Indeed, to produce an exact replica, I will have to construct my new shock absorber out of old parts, or artificially age the components I use so as to replicate the exact condition of the old shock absorber. Now, if my replacement of my car parts is of this sort, it is arguable (modulo doubts on a general theory of artefact identity) that my car will stay the same throughout the replacement of a great many of its parts (and let us ignore the question of how many a great many is). By contrast, if I am replacing old shock absorbers by new ones which are not modelled on the old ones, we find ourselves in a difficulty. Of course, it might be-to take the closest case to the one just discussed-that our new shock absorbers have been produced by the same production process that produced the original ones. They will thus be new tokens of the same type as the original. But, since they are not modelled on the original, the original does not, according to our previously defended position, survive in or as these new shock absorbers. Such a consideration is not, of course, fatal to the car's identity. A change in one or two components does not seem to affect the identity of the larger artefact, even where the original components have been replaced by ones of a different design. And 9

See Hirsch [2], ch. 7.

Concepts of the Person


in this case, what Parfit says seems to be true. The new parts, even if they do not match the originals, become part of the larger artefact simply by fitting into the overall structure of the larger item. So long as the process of replacement is gradual, we feel no violence done to our intuitions by the claun that the artefact has retained its identity (though this is still being said with due caution, in view of our general worries about artefact identity). Telecloning, on our account, is rather different from this last kind of case. For in telecloning, the original plays an important causal role in the production of the replica. It is because of the causal role, together with the satisfaction of our structure and matter conditions, that it seemed so natural to maintain that items can survive the teleclone. Now what Parfit has not mentioned in the case of the two operations is the causal role played by the original brain cells. But, in keeping with our earlier conclusions, it should now be clear that this is a matter of some importance. How, after all, is the surgeon to go about replacing my brain cells with non-defective replicas? Presumably, if I am to survive the process, then either there is a bigger difference between the two cases than Pamt mentions, or they are similar in respect of causality. First consider the latter. If I am my brain, then presumably certain structures in my brain will be distinctive of me, and would need to be preserved in order for me to be preserved. I t ~ay even be that certain structural features of my brain cells would need to be preserved in order for me to be preserved. Let us assume that this is so. Pamt, then, is assuming that the surgeon's intervention retainS' these structural features of my cells that are distinctive when putting in replicas of my cells. The replicas will keep those features of structure, and be in those structural relations to other cells, that are distinctive of me. But if this is how the case is meant to be understood, then there is no difference, as Parfit indeed says, between the two cases. But Nagel need make no difference between them either. For what happens in Case One and in Case Two is that my brain cells play a crucial causal role in the production of their replicas. Thus my brain cells, individually, survive as the new cells. And thus my brain survives as the new brain. Of course, it is an open question whether we should now associate this case with the case of the episodic hill in Chapter 4. In the latter case, the chain of surviving stages were all allocated to the one hill. And Nagel would certainly be able to say of


Concepts of the Person

both Partit's cases that what has happened is that my brain has persisted despite the changes in cells between stages. The survival of the damaged cells as their replicas would underpin this very identification. Let us suppose, for the sake of the argument, that Nagel did indeed take such a line. Then the claim that I am my brain does not need to be modified in the light of the Partit example. By contrast, suppose that, somehow or other, my existing brain cells play no causal role in the production of the non-defective cells inserted by the surgeon. This time we can imagine, perhaps, that new brain cells can be grown in some appropriate culture or medium, and that one health brain cell is pretty well like another. But now there would seem to be a real difficulty in understanding how I could survive both operations. Maybe what happens is like this. What makes me the person I am are certain brain structures and certain electrochemical states and changes in my brain. Any old brain cells could support such structures and so the gradual replacement of my brain cells does not threaten its identity. As each new cell is inserted, it gets organized into certain overall structural patterns, and receives appropriate chemical messages which in turn modify its responses to other electro-chemical signals. In this sort of case, it would be a matter of some importance whether my cells are replaced gradually or all at once. For, in order to be incorporated into the right structures and be responsive in an appropriate way to electro-chemical signals, new cells have to be associated with the existing arrangement of old ones. The sudden replacement of all my brain cells would involve the destruction of that brain, for it would involve destruction of the structures that were distinctive of it. In this case, then, Partit's two operations have very different consequences. Indeed, we would readily understand it if Nagel chose to maintain that I do persist through the one operation, but cease to exist given the other. In fact, these two cases are simply drawn from a whole spectrum of cases which could be considered. But, whatever kind of case we chose to discuss, much would depend on the copying question. Survival requires the participation of an original (stage, or item) in the causal process, that results in the existence of the item or stage in which the original survives. Of course we can argue about the relation of survival to identity, as we have been doing all through this book. But Partit's two cases do not pose the fatal dilemma for

Concepts of the Person


the brain theorist that he imagines. The brain theorist may, or may not, be committed to the view that identity is what matters. But what such a theorist has to be committed to is the view that the identity, or the survival, of the brain is what matters for a person's identity or survival. If anything, the attractiveness of the brain theory is increased by the way it responds to the cases imagined by Parfit.


In exploring various developments of the concept of a person, we have repeatedly come up against the same difficulty. Although a paticular development commends itself as plausible, we seem to be unable to marshal considerations that can decide for us whether this development is more useful or productive than any of its rivals. We do not know how to assess values the concept may have in these cases. But why should we be faced with such a difficulty? One reason for this may be that we have no idea how to set about testing one particular development against its rivals. Consider, by contrast, the case in the sciences. Although writers like Quine have suggested that there is no hard and fast theory-observation distinction, we regularly do make such a distinction. 1O Even if we admit, with Quine, that the boundary between them is not fixed, we can make some distinction between theory, as that which is set up for test, and observational data, as desiderata against which the theory is to be tested. But in the case of our theories of the person, we have no clear idea of what the body of testing data would be. We are dealing, after all, with intuitions, notions that we have in advance of all our conceptual analyis and development. Unless we can find some principled way of weighting these pre-theoretic intuitions, we can have no principled way of settling upon some as data against which our theory of the person is to be tested. Of course, we could settle, in some given case, for some intuitions to be held constant, so to speak, and use these as the testing-bed of theory. Indeed, this seems to be how philosophers often proceed. But these intuitions may them10 Hence his ability to treat Qrdinary particulars as posits of theory (in, e.g., Quine [2] ch. 1): see ch. 7, n. 8.


Concepts of the Person

selves be challenged by some other development of the concepts we are studying. In some cases, we may claim that general agreement settles which intuitions are the most central. But the literature alone on personal identity bears witness to the fact that there is no such agreement in the case of our concept of a person. We can point up the difficulty very clearly by reflecting on the role of the concept in moral philosophy. Locke puts the moral and legal role of persons in the following tenns: Where-ever a Man find what he calls himself, there I think another may say is the same Person. It is a Forensick Tenn appropriating Actions and their Merit; and so belongs only to intelligent Agents capable of a Law, and Happiness and Misery .... whatever past Actions it cannot reconcile or appropriate to that present self by consciousness, it can no more be concerned in, than if they have never been done: And to receive Pleasure or Pain; i.e. Reward or Punishment, on the account of any such Action, is all one, as to be made happy or miserable in its first being, without any demerit at all. For suppose a Man punish'd now, for what he had done in another Life, whereof he could be made to have no consciousness at all, what difference is there between that Punishment and being created miserable?1I

There seems to be a certain force in Locke's observations. Yet the difficulty with this forensic concept is knowing whether we should reject theories that pennit one person to be responsible for the deeds of another, but which make sense of Parfit-style survival, or whether, by contrast, we should take Locke's own passage here as developing a theory of the person which itself can be tested against our other views. We can think briefly, then, about what to say on the legal issues posed by the forensic concept when taken in conjunction with what we know about amnesia. Studies of various amnesias show that certain patients are genuinely afflicted with loss of memory. Is it right, then, to try such a patient for an offence which the patient cannot now recall committing (although there is psychological coherence, let us suppose, between the person committing the offence and the patient before us, as well as bodily continuity). If the law is not to be an ass, then it must make the tests for the presence of amnesia very stringent. And even Locke noticed that there may be such a difficulty here that amnesia is no more of an excuse than drunkenness. 11

Locke (1),



Concepts of the Person


Gibbensand Hall Williams give attention to the question of whether, in the case of an absolutely genuine amnesia, there would be a problem about the accused's fitness to plead. They point out: 'Many trials proceed in the absence of a witness whom the accused considers essential for his defence, or without the assistance of the evidence of the accused himself. Amnesia would therefore not appear to be relevant on this preliminary issue ... '12 But the Lockean would here object to the use of the comfortably vague term 'the accused'. If they mean that the accused may be the same person as did the criminal deed, although now genuinely amnesic with respect to that deed, then they are simply operating with a different concept of the person from Locke's one. They are begging the question against the Lockean. If, however, the issue of the accused's identity with the perpetrator of the criminal act is still open, and depends on the accused genuinely remembering doing the deed, then the problem is not one of fitness to plead. Rather, on Locke's view, the wrong person is being accused of the crime, for the person who did commit it no longer exists. It might now be argued that we were wrong to take this difficulty in the first instance as being a difficulty about theory and observation. An objector might want to say that the problem concerns, rather, the collision of two different theories of the person. Such an objection does not take us very far, even if we were to concede some of its force. For when two theories collide, we are able to check them for coherence with the rest of our body of theory, and also to test both of them against some shared observational evidence. After all, if there were no such shared observations or predictions common to both theories, what sense could we make of their rivalry? Applying this to the case in hand, we then have to take the forensic conception, and-let us say-the body conception, as two rival theories which we must test against our other other intuitions, taken as data. We also have to think about other areas of theorymoral theory, our theories about other objects and their identity, and so forth-with which we are to make our theory of the person cohere. But again, I would argue, the hope for progress is doubtful. Duhem's thesis that hypotheses can be indefinitely retained even in 12

In Whiny and Zangwill [1], p. 260.


Concepts of the Person

the face of adverse data hits us with a vengeance. 13 For we are operating in an area of great conceptual indetenninacy . We cannot look to our theories of the identity of other kinds of things, nor to our theory of morality, for the kind of definitive test that would settle the issue for or against the Lockean. Instead, as Quine has so forcefully argued in other contexts, we are faced with the possibility of making compensatory adjustments else where to accommodate whatever theory of the person we choose to fix on.14 The progress of science is slow. If I were a good Quinean, then I might be consoled by the thought that at least some of the theorizing in this book may itself have contributed to the development one day of a theory of the person within some appropriate branch of science. But anyone who is that much of a Quinean ought also to be sceptical enough to recognize that the indeterminacy in the concept of a person, like other conceptual indetenninacies, is liable to be around for some time. Early in my treatment, I pointed out that the notion of conceptual indetenninacy I am using is not meant to be the same as Quine's notion of translational indetenninacy or the associated notion of inscrutability of reference. 15 That point still stands. But the weaker thesis I have defended here, namely that our everyday notions about persons are vague, probably confused, possibly inconsistent, and capable of development in a variety of ways, is still alive and well. Persons, as we might have suspected all along, have proved to be somewhat more complicated objects than hills or lakes. So our modest success at developing an account of identity, based on survival, for other particulars has not carried over to the case of persons.

See Duhem [1] and Quine's repeated defence of this position. One of Quine's classic statements of this position is in 'Two Dogmas of Empiricism': 'The totality of our so-called knowledge or beliefs, from the most casual matters of geography and history to the profoundest laws of atomic physics or even of pure mathematics and logic, is a man-made fabric which impinges on experience only along the edges. A conflict with experience at the periphery occasions readjustments in the interior of the field. Truth values have to be redistributed over some of our statements. Reevaluation of some statements entails reevaluation of others, because of their logical interconections ... But the total field is so underdetermined by its boundary conditions, experience, that there is much lattitude of choice as to what statements to reevaluate in the light of any single contrary experience.' eQuine [1], pp. 42-3.) 15 See the title essay of Quine [7]. 13 14

Concepts of the Person



The present chapter opened by making one last attempt to develop a useful concept of the person along Parfit's lines. This time, a way was shown of making survival in our sense constitutive of personal identity. The enterprise was moderately sucessful, and can be applied to cases of equal and of unequal branching. However, a new difficulty with developing a concept of the person was then identified. According to it, there are different levels at which the concept can be developed. Our normal thinking about persons does not give any guidance about which is the preferred level of development. In combination with the type-token problem, this threatened to spell the end of conceptual development in this area. One way of tackling the new difficulty involved taking a materialist stand, and letting the identity of the person be associated with the identity of a brain. Although the materialist account brings us no nearer to solving the type-token problem it does survive what Parfit puts forward as a 'fatal' dilemma for it. A brief consideration of the forensic conception of the person again pointed up the difficulty faced by those engaged in conceptual development. This time, we have to notice the difficulty of deciding what counts as data against which to test our theory of the person. In an attempt to disarm one part of my negative pronouncements on the problem of persons, it might be argued that the type-token dualism afflicting our thinking about persons is no worse than the ambiguity of dimensions afflicting our normal consideration of particulars. Perhaps persons can be viewed under either guise, just as particulars can be viewed as three-, four-, or even fivedimensional. The difference, as I will argue in a moment, between type and token notions of the person is not akin to a difference in dimension. It may now seem hardly surprising that some theorists of the person are foundationalist or non-reductionist. Both body-based and connectedness-based theories looked promising, but have proved to have difficulties associated with them. If we return to Strawson's well-known argument that persons are essentially subjects both of predicates ascribing mental states and predicates ascribing physical states, this gets us out of some of our


Concepts of the Person

difficulties. 16 If the move is coupled, as it was in Strawson's original argument, with an appeal to the primitiveness of the concept of the person, then, as we have seen, the cost of the move is that it avoids the issue of saying in what the unity of a person consists. It strikes me, however, that the case of persons is rather different from the earlier puzzle case involving artefacts. It was clear that certain very general difficulties make it unlikely that we could ever come up with a decent theory of artefact identity, even when applying the apparatus of survival theory. The chain theory will, certainly, give us some account of the unity of artefac¥, but the relation of structure to function was too loose to enable it to apply to the cases where it was most needed-those where dispute is possible. However intractable the case of artefacts, there is not a similar problem in the case of persons. For it is now clear that the chain theory can work on either a body-based account (where this account includes suitable reference to brains) or on a connectedness-based account. Our problem is in determining whether persons are best thought of in one ~r other of these ways. Taking the latter account first, it has been shown that neither amnesia, nor ownership, are real problems for the psychological account of unity. Rather, the real problem is that no psychological account will apparently do justice to our concept of persons as individual, body-based 'tokens'. Although Parfit and others have tried to show that our concern for bodies is not a concern for what is important, it is wrong, in my view, to dismiss our body-mindedness as mere bias. Rather, a good psychological theory will either find some way of doing justice to this aspect of our thinking or reveal clearly why it should be abandoned. As far as I can see, natural though the connectedness account is, it fails to deal adequately with this pre-theoretic attitude. Even attaching a connectedness-based account to one part of the body-the brain-and arguing that we then have a rapprochement between materialism and psychological considerations does not do the trick. It is tempting to see a whole chain of reductions taking place here. First, the high-level psychological connections mentioned by Parfit are reduced to the interaction of variously structured mechanisms. These mental devices are reduced in turn 16

See n. 33 to Ch. 8.

Concepts of the Person


to interacting complexes of simpler mechanisms until we reach a point where we are comfortable about identifying the psychological with the physical. If I am still the same person I was ten years ago, then presumably my present states are connected to a suitably high degree with the states I had then. If we go along with the reduction, then this means that there is some appropriate causal relation between my brain structures now and my brain structures then-a relation that supports this psychological unity. As we have seen, however, concern for, and care about, brains is no part of our usual concern for someone thought of as an individual, embodied token. But to go all the way and identify the person with the body (including brain, heart, liver, endocrine system, and everything else) ignores the plain fact that it is the doings, the experiencings, the desirings and strivings that matter to us just as much as the body which supports these things. Nor is it just the case that we value the actual doings as much as we value the body that does them. Recall, for a moment, the five-dimensional view of particulars. There is a double potential for persons, viewed as being either tokens or types, hardware or software. I could be the same, ina sense, while yet my body is (within limits) different. My body could be the same while yet my behaviour, styles of response, loves, and hates are (within limits) different. There is, after all, a sad truth in the observation that sometimes we love people for what they might have been as much as for what they are. The 'might have been' for persons ranges over the two domains which are at the root of our problem. There are yet further complications in the story about persons that have not even been mentioned. We might wish to talk of persons as systems in which certain production processes, recipes for behaviour, interact with the behaviour produced. But this would be a feeble attempt at giving the ecology of persons. For it would ignore two things, already familiar in some philosophical traditions, but ignored in recent analytical work. Dewey, for example, would protest that we spend too little time in explaining how the individuality of a person is at least in part a function of involvements with others.; for action involves location within a community of persons, and individuation requires different involvements in different groups. He would also remind us of the need to take account of growth-the process which involves the


Concepts of the Person

continuous interpretation and reconstruction of experience. 17 Any account of psychological connectedness informed by Dewey's insights would allow that my view of what I was doing then is to some extent coloured by the position I am in now. Such considerations are familiar to students of literature, but the subtleties of a Proust, for example, are ignored by the thin accounts of memory and experience given in books like this. So if we were tempted to explore the ecological dimension of personal unity, we would have to give due weight to the dynamics of the person, and the relation of the person to other persons and the rest of the environment. An exploration along these lines might well model itself on the ecology of daffodils; for they too contain recipes for behaviour-the pushing up of shoots and bursting into annual bloom-and are affected by their environments. But I am personally sceptical about the prospects for any such investigation. The dual nature of our conception of the person is, I suspect, likely to be as great a stumbling-block for the systems theorist as for the hardy analyst. Before succumbing to a sense of failure in the face of these reflections, they can be put into perspective by considering the gains that have been made so far. Although we do not have a comprehensive theory of unity for every conceivable sort of thing, a general framework is available for approaching unity or identity questions. Take, for example, a case not so far considered. Biologists commonly talk of species as staying the same through countless generations. In other cases, however, a current species is classed simply as a descendant of an earlier one. It looks 17 The central theme of the early chapters of Dewey [1] is that growth, or self-renewal, involves living things in interaction with their environment. The environment itself is what promotes, stimulates, inhibits, or hinders the characteristic activities of living things. For human beings, a social environment is an indispensible condition of the realization of the individual's tendencies. Famously, Dewey's conception of growth is linked to panicu1ar development he makes of the concept of education. The school, according to his theory, should be a. special growth-promoting environment, which simplifies the complexity of the wider social environment, eliminates less wonhy features of it, and overcomes limitations of the social groups into which individual persons were born. The growth and reconstruction associated with living and with education go on throughout each person's life, according to Dewey. Moreover, if his association of types of life and growth with the nature of democratic society is correct, it would· follow that different social organizations would produce different kinds of persons. For more on Dewey, and on conceptual development in eduction, see my 'Analysis, Development and Education' in the British Journal of Educational Studies, 24 (1986).

Concepts of the Person


overhwelmingly plausible to suggest that we can best regiment discussion of the cases involved by thinking of the matter in terms of the survival of species. To do so, we would have to investigate just what constitutes the matter and structure of a species-an issue that the modem systematists have already studied. For them, a species is a superindividual, a breeding population, whose components are the particular plants or animals. 18 Then we would need to consider the nature of causal connections between stages of a species, to try to distinguish these cases where there is survival of a high enough degree for unity from those where there is survival without identity. The applicability of the conditions to this case is not surprising. After making a case early on for the utility of the general framework, its subsequent application to the problem of discontinuity and to the quite separate problem of amnesia and branching amply confirms its value. That the general framework of survival theory stands in close relations to plausible psychological models of how we latch on to structures in the world gives it, in my view, further credibility. The rejection of what was called the 'extensionalist myth' thus seems well founded, even if, in these matters, there is no possibility of conclusive refutation. A number of other results have been established, including some that discredit well-known strategies in metaphysics and the theory of identity. Spatio-temporal continuity has been shown to be fairly worthless in giving an account of the identity of anything; given the difficulties in making precise just what our intuitive notion of continuity is meant to embrace, this result is fortunate. In the case of the concept of sortal, I have argued that it is less primitive than the concept of structure, and can mislead us if used in the theory of survival. Survival, after all, can be a relation between different sorts of things. Our recognition of survival is thus more closely in touch 18 The conception of species as individuals is a response of sons to Ghiselin and Hull's arguments that species are not natural kinds. For discussion and references see Sober [1]., The case of species is just one of many that might have been mentioned. For other examples, consider the claims that can be made about societies, or cultures. Like species, cultures can develop, change, grow, decline, or thrive without losing their identity. Only by examining these sons of cases in detail could we get to grips with the Eusa problem and produce a decent account of the metric of change. For in all cases, retention of identity, I suspect, will depend on there not being too much change in matter and structure between stages remote in time. Apan from the denial that limits to change are sortal relative, I am conscious of not having suggested so far any positive theory of change.


Concepts of the Person

with the genuine similarities and differences in the world than our recognition of sorts. It is against this backgroWld, I think, that we should view our problems about personal identity. No doubt it would have been pleasant to find some overwhelming support for the 'token' theory of the person, for this comes closest to expressing my own pretheoretic beliefs. But to adopt this accoWlt would be impossible Wlless I were to close my eyes to the persuasive arguments that support the 'type' theory. It is necessary, then, to recognize virtues in both accoWlts, and to voice, therefore, the suspicion that we are systematically ambiguous in our references to, and our thinking about, persons. The ambiguity isolated may simply be one of many that afflicts our reflections Qn this topic. Considering the matter thus in terms of ambiguity might give us some hope that the situation is not so bad as I have been making out. After all, it was argued in Chapter 5 that particulars are not 'really' three-, or four-, or even five-dimensional. So might it not be suggested that the ambiguity about persons is no worse than this: persons are not really tokens or types, but can be viewed now in this, and now in that, way. The ambiguities, however, are not similar, and even this last hope of comfort has to be set aside. It is entirely Wlderstandable that we do not switch vocabulary when switching our attention from one dimension to another. So any exploration of different dimensions is bOWld to bring in its train the kind of ambiguity isolated in the fifth chapter. But nothing similar to switching dimensions is taking place when we think about persons. Rather, if my diagnosis has been correct, there is something of which we are not normally aware when we refer to persons. This is that we bring two very different, and irreconcilable, kinds of assessment to bear when we consider the question of their Wlity. To put the disanalogy between the two kinds of case in a crude way, we might say that the point about dimensions applies to items that are one and the same, whatever level we are using to describe them. A particular like my desk is the same particular whether I am describing it in three-, four-, or five-dimensional terms. The ambiguity in our thinking about desks is thus not an ambiguity about the number of objects being described: there is not a threedimensional desk in my office sharing space--but not time-with a four-dimensional particular. By contrast, our difficulty with the

Concepts of the Person


mode of describing and responding to persons is precisely a difficulty about the object of our attention. If recipes and behaviour make a person, then a person is a different sort of thing from an individual token. If I am right about our different modes of response to persons, then the situation is like one where two different things compete for our attention, evaluation, and care. When we try to focus our analytical gaze on persons we cannot escape a kind of double vision. At the risk of emphasising the obvious, let me repeat that the duality in our view of persons discussed here and at the end of Chapter 9 is not the familiar duality of the mental and the physical, though it is easily confused with it. The discovery of this duality is, to my mind, one of the most important outcomes of the discussion in the last three chapters. The result is not just a limitation on survival theory, but a problem for any account of persons, however reductionist or non-reductionist it is. It should also be clear by now that my own commitment to reductionism is not of great eliminative significance. My own view on the issue of reductionism is quite simple. A complex artefact is a collection of parts, but has properties that its individual parts lack. None the less, we can explain the properties of the whole item by reference to properties of its parts. A scientific account of the powers and capacities of whole natural objects will likewise explain what their features consist in by appeal to the features of the parts and their relations to each other. Just so is it, in my view, with the property of unity or identity. In saying that the unity of a particular consists in certain causal relations among its spatial or spatio-temporal parts, we only in a sense reduce that particular to its parts. To insist that a broad particular is no more than a sum of spatiotemporal parts would be like maintaining that a complex, structured item is simply an aggregate of matter. Certainly there is a limited way in which this is true. One problem with taking the latter line, as we have seen, is that things like statues become indistinguishable from mere lumps of clay. This would be reductionism with a vengeance, and would involve an implausible impoverishment in our descriptions of the universe to which I have no commitment. But just as a statue is clay with structure, so a particular thought of in terms of its spatio-temporal extent is a sequence of stages with structure-this structure being defined at


Concepts of the Person

least in part by the survival relation as I have described it. No doubt this crude way with the issue of reduction ism leaves many questions unanswered, but these can be left for exploration in other places. If I were now to start on the project of the book, knowing what I now know, and armed with the discoveries, hunches, and arguments that are scattered through the preceding pages, the result-I hope-would be a clearer, more focused account of problems involving identity. However, the same major themes would be stressed in much the same order. The claims of the survival account over any elucidation in terms of spatio-temporal continuity, the dimensional ambiguity of particulars, the theory of real possibility, the rejection of the extensionalist myth, the denial of the importance of memory to self-identity, and the seemingly irresolvable duality in our thinking about persons as types and persons as tokens-these would all have a central place in any revised version. As it is, any reader stimulated by what is written in the present version will, I hope, be able to approach the study of identity aware of some of the pitfalls to be avoided and of the problems yet to be solved.

Bibliography The bibliography, although selective, is meant to be relatively comprehensive and lists several works that are not cited in the main body of the book. Ackermann, R. [1]. 'Sortal Predicates and Confirmation', Philosophical Studies, 20 (1969). Adams, R. [1]. 'Primitive Thisness and Primitive Identity', Journal of Philosophy, 76 (1979). Albrinon, Rogers [1]. 'On Wingenstein's Use of the Term "Criterion"', in Pitcher [1]. Ames, L. L., jun. [1]. 'The Genesis of Carbon Apatite', Economic Geology, 54 (1959). Anderson, A. R., and Be1nap, N. D. [1]. Entailment (Princeton: Princeton University Press, 1975). Anscombe, G. E. M. [1]. 'The Principle of Individuation', Proceedings of the Aristotelian Society, Suppl. Vol. 27 (1953). Aristotle [1]. The Works of Aristotle, ed. W. D. Ross (Oxford: Oxford University Press, 1929-52). Aschkenasy, Jeannie, and Odom, R. D. [1]. 'Classification and Perceptual Development', Journal of Experimental Child Psychology, 34 (1982). Ayer, A. J. [1]. Philosophical Essays [London: Macmillan, 1954). - - [2]. The Problem of Knowledge (Harmondsworth: Penguin, 1956). - - [3]. The Concept of a Person, and Other Essays (London: Macmillan, 1964). - - [4]. 'Identity and Reference', Philosophia, 5 (1975). Ayers, Michael R. [1]. 'Individuals Without Sortals', Canadian Journal of Philosophy, 4 (1974). Barcan (Marcus), Ruth [1]. 'The Identity of Individuals in a Strict Functional Calculus of Second Order', Journal of Symbolic Logic, 12 (1947). - - [2]. 'Interpreting Quantification', Inquiry, 5 (1962). - - [3]. 'Essential Anribution', Journal of Philosophy, 68 (1971). Benacerraf, Paul [1]. 'What Numbers Could Not Be' , Philosophical Review, 74 (1965). Bennett, Daniel [1]. 'Essential Properties' ,Journal of Philosophy, 66 (1969). Bennett, J. [1]. Kant's Analytic (Cambridge: Cambridge University Press, 1966). Bertalanffy, L. von, and Rapoport, A. (eds.) [1]. General Systems Theory (Ann Arbor: 1962).



Black, Max [1]. 'The Identity of Indiscemibles', Mind, 61 (1952); reprinted in Loux [1]. Borst, C. V. (ed.) [1]. The Mind-Brain Identity Theory (London: Macmillan, 1970). Bradley, M. C. [1]. Critical Notice of Wiggins, Identity and SpatioTemporal Continuity, in Australasian Journal of Philosophy, 47 (1969), 69-79. Brennan, Andrew [1]. 'Persons and Their Brains', Analysis, 30 (1969). - - [2]. 'Personal Identity and Personal Survival', Analysis, 41 (1982). - - [3]. 'Survival', Synthese, 59 (1984). - - [4]. Critical Study ofE. Hirsch, The Concept of Identity, in Nous, 18 (1984). - - [5]. 'Amnesia and Psychological Continuity', in Copp, R. and McIntosh, J. J., New Essays in the Philosophy of Mind (Canadian Journal of Philosophy Suppl. Vol. 11, 1985). Brody, Baruch [1]. 'Locke on the Identity of Persons', American Philosophical Quarterly, 9 (1972). - - [2]. 'Why Settle for Anything Less Than Good Old-Fashioned Aristotelian Essentialism?', Nous, 7 (1973). - - [3]. Identity and Essence (Princeton: Princeton University Press, 1980). Burge, Tyler (1]. 'A Theory of Aggregates', Nous, II (1977). - - [2]. 'Belief De Re', Journal of Philosophy, 74 (1977). - - [3]. 'Mass Terms, Count Nouns and Change', Synthese, 31 (1975). Butler, J. [1]. 'Of Personal identity' (first dissertation to The Analogy of Religion), printed in Flew [3] and in Perry [5]. Care, N. and Grimm, R. H. [1]. Perception and Personal Identity (Cleveland: Ohio University Press, 1969). Carnap, R. [1]. Meaning and Necessity (2nd edn.; Chicago: Chicago University Press, 1956). Carter, W. R. [1]. 'On Contingent Identity and Temporal Worms', Philosophical Studies, 41 [1982] 213-30. Cartwright, Helen M. [1]. 'Quantities', Philosophical Review, 79 (1970). - - [2]. 'Heraclitus and the Bath Water', Philosophical Review, 74 (1965). - - [3]. Review of Wiggins. Sameness and Substance, in Philosophical Review, 92 (1982). Cartwright, Richard [1]. 'Identity and Substitutivity', in Munitz [1]. - - [2]. 'Scattered Objects'. in Lehrer, K. (ed.), Analysis and Metaphysics (Dordrecht: D. Reidel, 1975). - - [3]. 'Indiscemibility Principles', Midwest Studies in Philosophy, 4 (1979). Castaiieda, Hector-Neri [1]. 'Identity and Sameness', Philosophia, 5 (19~5).



Chandler, H. S. [I). 'Wiggins on Identity', Analysis, 29 (1968/9). - - [2]. 'Constitutivity and Identity', Nous, 5 (1971). Chappell, v. c. [I). 'Sameness and Change', Philosophical Review, 69 (1960). - - [2). 'Stuff and Things', Proceedings of the Aristotelian Society, 71 (1970/1). Chisholm, R. M. [1]. 'Identity Through Possible Worlds', Nom, 1 [1967). - - [2). 'The Loose and Popular and the Strict and Philosophical Senses ofIdentity', in Care and Grimm [I). - - [3). 'Parts as Essential to their Wholes', Review of Metaphysics, 26 (1972/3). - - [4]. 'Problems of Identity', in Munitz [I). - - [5]. Person and Object (London: Alien & Unwin, 1976). Clark, E. V. [1]. 'What's in a Word?' in T. E. Moore (ed.), Cognitive Development and the Acquisition of Language (New York: Academic Press, 1973). Clarke, D. S. [1). 'Mass Terms as SUbjects', Philosophical Studies, 21 (1970). Cornman, James W. [I). 'Types, Categories, and Nonsense', in Rescher [I). - - [2). 'The Identity of Mind and Body', in Borst [I). - - [3). Materialism and Sensations (New Haven: Yale University Press, 1971). Darwin, Charles [I). The Origin of Species (London: Murray, 1900). Davidson, D., and Hintikka, J. (eds.) [1 ). Words and Objections (Dordrecht: D. Reidel, 1969). Dawkins, R. [I). The Selfish Gene (Oxford: Oxford University Press, 1976). Dennett, D. [I). Brainstorms (Hassocks, Sussex: Harvester, 1979). Descartes, R. [I). Philosophical Works, tr. Haldane E. S. and Ross G. R. T. (Cambridge: Cambridge University Press, 1969). Dewey, John [I). Democracy and Eduction (New York: Macmillan, 1916). Duhem, P. [I). The Aim and Structure. of Physical Theory (Princeton: Princeton University Press, 1954). Dummett, M. A. E. [1). Frege: Philosophy of Language (London: Duckworth, 1973). - - [2J. 'Does Quantification Involve Identity?', in Lewis, Harry, and Saarinen, E. (eds.), Peter Geach: A Profile (Dordrecht: Reidel, forthcoming). Elliot, R. [I). 'How to Travel Faster than Light', Analysis, 41 [1981). - - [2). 'Going Nowhere Fast?', Analysis, 42 [1982). En~, Berent [1]. 'Numerical Identity and Objecthood', Mind, 84 (1975). Evans, Gareth [1). 'Identity and Predication', Journal of Philosophy, 72 (1975).



- - [2]. 'The Causal Theory of Names', in Schwartz [1]. - - [3]. 'Vague Objects', Analysis, 39 (1978). - - [4]. The Varieties of Reference, ed. John McDowell (Oxford: Clarendon Press, 1982). Flew, A. [1]. 'Locke and the Problem of Personal Identity', Philosophy, 26 (1951). - - [2]. ' "The Soul" of Mr A. M. Quinton', Journal of Philosophy, 60 [1963]. - - (ed.) [3]. Body, Mind and Death, (New York: Macmillan, 1964). Fodor, Jerry [1]. Psychological Explanation (New York: Random House, 1968). Forbes, G. [1]. 'Origin and Identity', Philosophical Studies, 37 (1980). - - [2]. 'Thisness and Vagueness', Synthese, 54 (1983). - - [3]. The Metaphysics of Modality (Oxford: Clarendon Press, 1985). Fraassen, Bas C. Van [1]. 'Essence and Existence', American Philosophical Quarterly Monograph (1977-l3). Frege, Gottlob [1]. The Foundations of Arithmetic, tr. J. L. Austin (Oxford: Blackwell, 1950). - - [2]. Grundgesetze der Arithmetik (Jena: Verlag Hermann Pohle, 1883, 1903), partly translated as The Basic Laws of Arithmetic, with introduction, by Montgomery Furth (Berkeley: University of California Press, 1967). Furth, Montgomery (1]. 'Transtemporal Stability in Aristotelian Substances', Journal of Philosophy, 75 (1978). Gale, R. M. (ed.) [1]. The Philosophy of Time (Hassocks, Sussex: Harvester, 1978). - - [2]. Review of Eli Hirsch, The Concept of Identity, in Journal of Philosophy, 80 (1983). Garner, W. R. [1]. Uncertainty and Structure as Psychological Concepts (New York: John Wiley & Sons, 1962). - - [2]. The Processing of Information and Structure (Potomac: Laurence Erlbaum, 1974). Geach, P. T. [1]. Reference and Generality (Ithaca: Cornell University Press, 1962). - - [2]. 'Some Problems About Time', Proceedings of the British Academy, 1965. - - [3]. Logic Matters (Oxford: Blackwell, 1972). - - [4]. 'Ontological Relativity and Relative IdentitY', in Munitz [2]. - - [5]. 'Names and Identity', in Guttenplan [1]. Geach. P. T., and Black, Max (eds.) [1]. Translationsfrom the Philosophical Writings of Gottlob Frege (Oxford: Basil Blackwell, 1952). Gibbard, A. [1]. 'Contingent Identity', Journal of Philosophical Logic, 4 [1975].



Gibson, J. J. [1]. The Ecological Approach to Visual Perception (Boston: Houghton MifHin, 1979). Goldstein, L. [1]. 'Quotation of Types and Other Types of Quotation', Analysis, 44 (1984). Goodman, N. [1]. Fact, Fiction and Forecast (New York: Bobbs-Merrill, 1965). - - [2]. The Structure of Appearance (New York: Bobbs-Merrill; 2nd edn., 1966). - - [3]. Languages of Art (New York: Bobbs-Merrill, 1968). - - [4]. Problems and Projects (Indianapolis: Bobbs-Merrill, 1972), Pt. IV. Grice, H. P. [1]. 'Vacuous Names', in Davidson and Hintikka [1]. Griffin, Nicholas [1]. 'Ayers on Relative Identity', Canadian Journal of Philosophy, 6 (1976). - - [2]. Relative Identity (Oxford: Clarendon Press, 1977). - - , and Routley, R. [1]. 'Towards a Logic of Relative Identity', Logique et Analyse (1979). Guttenplan, S. (ed.) [1]. Mind and Language (Oxford: Clarendon Press, 1976). Hacker, P. M. S. [1]. 'Substance: The Constitution of Reality', Midwest Studies in Philosophy, 4 (1979). Hacking, Ian [1]. 'On the Reality of Existence and Identity', Canadian Journal of Philosophy, 8 (1978). Hampshire, Stuart [1]. 'Identification and Existence', in Lewis, H. D. (ed.), Contemporary British Philosophy (London: AlIen & Unwin, 1956), 3rd Ser. Hempel, Carl G. [11. Aspects of Scientific Explanation (New York: Collier Macmillan, 1965). Hirsch, Eli [1]. 'A Sense of Unity', Journal of Philosophy, 75 (1978). - - [2]. The Concept of Identity (Oxford: Clarendon Press, 1982). Hobbes, Thomas [1]. De Corpore, in Molesworth, William (ed.), The English Works of Thomas Hobbes (London: John Bohn, 1839-45), vol. i. Hofstadter, D. R. and Dennen, D. [1]. The Mind's I (Brighton: Harvester, 1981). Hull, David [1). 'Units of Evolution : A Metaphysical Essay', in JenseD and Harre [1). Hwne, David [1). A Treatise of Human Nature, 1739 (Oxford: ClarendoD Press, 1978). - - [2). An Enquiry Concerning Human Understanding, 1748 (Indianapolis: Bobbs Merrill, 1955). Ishlguro, H. [1]. Leibniz's Philosophy of Logic and Language (London: Duckworth, 1972). Iversen, S. D. [1]. 'Temporal Lobe Amnesia', in Whitty and Zangwill [1).



Jensen, U. J., and Harre, R. (eds.) [1]. The Philosophy of Evolution (Hassocks, Sussex: Harvester, 1981). Kant, Imrnanuel [1]. Critique of Pure Reason, tr. Nonnan Kemp Smith (London: Macmillan, 1964). - - [2]. Prolegomena to Any Future Metaphysics, 1783; tt. P. G. Lucas (Manchester: Manchester University Press, 1953). Kaplan, David [1]. 'Transworld Heir Lines', in Loux [2]. Katz, Bemard D. [1]. 'The Identity of Indiscernibles Revisited', Philosophical Studies, 44 (1983). Kemler, Deborah [1]. 'Exploring and Re-exploring Issues of Integrality, Perceptual Sensitivity and Dimensional Salience' ,Journal of Experimental Child Psychology, 36 (1983). Kirwan, Christopher [1]. 'How Strong are the Objections to Essence?', Proceedings of the Aristotelian Society, 71 (1970/1). Kitcher, Patricia [1]. 'Kant on Self Identity', Philosophical Review, 91 (1982). Kofika, Kurt [1]. Principles of Gestalt Psychology (New York: Harcourt, Brace and World, 1935). Krecz, C. A. [1]. 'Parts and Pieces', Philosophy and Phenomenological Research 46 (1986). Kripke, S. [1]. 'Identity and Necessity', in Munitz [1]. - - [2]. Naming and Necessity, rev. edn. (Oxford: Blackwell, 1980). Leibniz, G. W. [1]. Discourse on Metaphysics, translated by P. Lucas and L. Grint (Manchester: Manchester University Press, 1952). - - [2]. Philosophical Papers and Letters, tt. and ed. Leroy Loemker (Dordrecht: D. Reidel, 1969). Lewis, David [1]. 'Counterpart Theory and Quantified Modal Logic', Journal of Philosophy, 65 (1968), reprinted in Loux [2]. - - [2]. Counterfactuals (Oxford: Basil Blackwell, 1973). - - [3]. 'Survival and Identity', in Rorty [1]; reprinted with postscript in Lewis [4]. - - [4]. Philosophical Papers (Oxford: Oxford University Press, 1983). - - [5]. 'Individuation by Acquaintance and by Stipulation', Philosophical Review, 92 (1983). - - [6]. On the Plurality of Worlds (Oxford: Basil Blackwell, 1986). Linsky, Leonard [1]. Critical Notice of Geach, Reference and Generality, in Mind, 73 (1964). - - [2]. Referring (London: Routledge and Kegan Paul, 1967). - - [3]. Names and Descriptions (Chicago: University of Chicago Press, 1977). Locke, John [1]. An Essay Concerning Human Understanding, 2nd ed., 1694; ed. P. Nidditch (Oxford: Clarendon Press, 1975). Lockwood, Michael [1]. 'Identity and Reference', in Munitz [1].



Loux, M. J. (ed.) [1]. Universals and Particulars (Garden City: Anchor Books, 1970). - - (ed.) [2]. The Possible and the Actual (Ithaca: Comell University Press, 1979). Lowe, E. J. [1]. 'On the Identity of Artifacts', Journal of Philosophy, 80 (1983). - - [2]. 'Instantiation, Identity and Constitution', Philosophical Studies, 44 (1983). Lucas, J. R. [1]. A Treatise on Time and Space (London: Methuen, 1973). Luschei, Eugene C. [1]. The Logical Systems of Lesniewski (Amsterdam: North-Holland, 1962). Lycan, W. Gregory [1]. 'Noninductive Evidence: Recent Work on Wittgenstein's "Criteria" ',American Philosophical Quarterly, 8 (1971). McCabe, V. [1]. 'The Direct Perception of Universals', Synthese, 52 (1982). MacDonald, G. F. (ed.) [1]. Perception and Identity (Ithaca: Comell University Press, 1979). MacLeod, R. B., and Pick, H. L., jun. [1]. Perception: Essays in HonorofJ. J. Gibson (Ithaca: Comell University Press, 1974). Madell, G. [1]. The Identity of the Self (Edinburgh: Edinburgh University Press, 1981). Margolis, Joseph [1). 'Dracula the Man: An Essay in the Logic of Individuation', International Philosophical Quarterly, 4 (1964). - - [2]. 'Nature, Culture and Persons', Theory and Decision, 13 (1983). Martin, C. B., and Deutscher, M. [1]. 'Remembering', Philosophical Review, 78 (1969). Melior, H. [1]. 'On Things and Causes in Spacetime', British Journal for the Philosophy of Science, 31 (1980). - - [2]. Real Time (Cambridge: Cambridge University Press, 1981). Mendelsohn, Richard L. [1]. 'Rigid Designation and Informative Identity Sentences', Midwest Studies in Philosophy, 4 (1979). - - [2]. 'Frege's Begriffsschrift Theory ondentity' ,Journal of the History of Philosophy, 20 (1982). Millar, Alan [1]. 'What's in Look?', Proceedings of the Aristotelian Society, 86 (1985--6). Moore, G. E. [1]. 'Identity', in The Commonplace Book ofG. E. Moore, ed. Casimir Lewy (London: George AlIen & Unwin, 1962). Moravcsik, J. M. E. [1]. 'The Discemibility of Identicals'. Journal of Philosophy, 73 (1976). Morris, T. V. [1]. Understanding Identity Statements (Aberdeen: Aberdeen University Press, 1985). Munitz, Milton K. (ed.) [1]. Identity and Individuation (New York: New York University Press, 1971).



- - (ed.) [2]. Logic and Ontology (New York: New York University Press, 1973). Nagel, Thomas [1]. 'Brain Bisection and the Unity of Consciousness', Synthese, 22 (1971) and in Nagel [2]. - - [2]. Mortal Questions (Cambridge: Cambridge University Press, 1979). Nerlich, G. [1]. 'What Can Geometry Explain?', British Journal for the Philosophy of Science, 30 (1979). Noonan, H. [1]. 'Wiggins on Identity', Mind, 82 (1973). - - [2]. 'Sortal Concepts and Identity', Mind, 87 (1978). - - [3]. Objects and Identity (The Hague: Martinus Nijhoff, 1980). - - [4]. 'The Necessity of Origin', Mind, 92 (1983). - - [5]. 'Wiggins, Artefact Identity and "Best Candidate" Theories', Analysis, 45 (1985). - - [6]. 'The Only x andy Principle', Analysis, 45 (1985). - - [7]. 'A Note on Temporal Parts', Analysis, 45 (1985). - - [8]. 'The Closest Continuer Theory of Identity', Inquiry, 28 (1985). Nozick, R. [1]. Philosophical Explanations (Oxford: Clarendon Press, 1981). O'Connor, D. J. [1]. 'On Resemblance', Proceedings of the Aristotelian Society, 46 (1945/6). - - [2]. 'The Identity of Indiscemibles', Analysis [1954]; reprinted in Loux (1]. Odegard, Douglas [1]. 'Identity Through Time', American Philosophical Quarterly, 9 (1972). Parfit, Derek [1]. 'Personal Identity', Philosophical Review, 80 (1971). - - [2]. 'On "The Importance of Self-Identity" ',Journal of Philosophy, 68 [1971]. - - [3]. 'Personal Identity and Rationality', Synthese, 53 [1982]. - - [4]. Reasons and Persons (Oxford: Clarendon Press, 1984). Peacocke, C. [1]. 'Are Vague Predicates Incoherent?', Synthese, 46 (1981). - - [2]. Sense and Content (Oxford: Clarendon Press, 1983). Peirce, C. S. [1]. Collected Papers, ed. C. Hartshome, P. Weiss, and A. W. Burks (Cambridge, Mass.: Harvard University Press, 1933). Penelhum, Terence [1]. 'Hume on Personal Identity', Philosophical Review, 64 (1955). - - [2]. Survival and Disembodied Existence (London: Routledge & Kegan Paul, 1970). Perry, John [1]. 'The Same F', Philosophical Review, 79 (1970). - - [2]. Review of Wiggins, Identity and Spatio-Temporal Continuity, in Journal of Symbolic Logic, 35 (1970). - - [3]. 'Can the Self Divide?', Journal of Philosophy, 69 [1972]. - - [4]. 'Personal Identity, Memory, and the Problem of Circularity', in Perry [5].



- - (ed.) [5]. Personal Identity (Berkeley and Los Angeles: University of California Press, 1975). - - [6]. 'The Problem of Personal Identity', in Perry [5]. - - [7]. 'The Importance of Being Identical', in Rorty [1]. - - [8]. Review ofWilliams, Problems of the Self,Journal of Philosophy, 73 (1976). - - [9]. 'The Essential Indexical', Nous, 13 (1979). Piercy, M. F. [1]. 'Experimental Studies of the Organic Amnesic Syndrome', in Whitty and Zangwill [1]. Pitcher, G. (ed.) [1]. Wittgenstein; The Philosophical Investigations (London: Macmillan 1968). Plantinga, Alvin [1]. The Nature of Necessity (Oxford: Clarendon Press, 1974). - - [2]. 'World and Essence', in Loux [1]. - - [3]. 'The Boethian Compromise', American Philosophical Quarterly, 15 (1978). Price, M. [1]. 'Identity Through Time', Journal of Philosophy, 74 (1977). Prior, A. N. [1]. 'Report on the Analysis "Problem" No. 11', Analysis, 17 (1957), 122-3. - - [2]. 'Opposite Number', Review of Metaphysics, 11 (1957/8). - - [3]. Time and Modality (Oxford: Clarendon Press, 1957). - - [4]. Formal Logic (Oxford: Oxford University Press; 2nd edn; 1962). - - [5]. 'Time, Existence and Identity', in Papers on Time and Tense (Oxford: Clarendon Press, 1968). - - [6]. 'Things and Stuff', in Papers in Logic and Ethics (London: Duckworth, 1976). Puccetti, R. [1]. 'Brain Transplantation and Personal Identity', Analysis, 29 (1968). Putnam, Hilary [1]. 'It Ain't Necessarily So', Journal of Philosophy, 59 [1962]. - - [2]. 'Is Semantics Possible?', Metaphilosophy, 3 (1970). - - [3]. 'Meaning and Reference', Journal of Philosophy, 70 (1973). Quine, W. V. [1]. From a Logical Point of View (New York: Harper Torchbooks; 2nd edn., 1961). - - [2]. Word and Object (Cambridge, Mass.: MIT Press, 1960). - - [3]. Set Theory and its Logic (Cambridge, Mass.: Harvard University Press, 1963). - - [4]. Review of Geach, Reference and Generality, in Philosophical Review, 73 (1964). - - [5]. The Ways of Paradox (New York: Random House, 1966). - - [6]. 'Natural Kinds', in Rescher, N. (ed.), Essays in Honour of c. G. Hempel (Dordrecht: Reidel, 1969) and in Quine [7]. - - [7]. Ontological Relativity and Other Essays (New York: Columbia University Press, 1969).



- - [8]. Review of Munitz, Milton K. (ed.), Identity and Individuation, in Journal of Philosophy, 69 (1972). - - [9]. The Roots of Reference (La Salle: Open Court, 1974). - - [10]. Theories and Things (Cambridge, Mass.: Belknap Press, 1981). Quinton, A. M. [1]. 'The Soul', Journal of Philosophy, 59 (1962]. - - [2]. The Nature of Things (London: Routledge & Kegan Paul, 1972). Ramsey, Frank P. [1]. The Foundations of Mathematics and Other Logical Essays, ed. R. B. Braithwaite (Totowa, NJ: Littlefield, Adams, 1965). Reid, Thomas [1]. Essays on the Intellectual Powers of Man (1784), Ch. 4, 'Of Memory', reprinted in Perry [5]. Rescher, Nicholas (ed.) [1]. Studies in Logical Theory (Oxford: Basil Blackwell, 1968). - - , and Urquhart, Alasdair [1]. Temporal Logic (New York: Springer Verlag, 1971). Roberts, M. [1]. 'Lewis's Theory of Personal Identity', Australasian Journal of Philosophy, 61 [1983]. Robinson, D. [1]. 'Reidentifying Maner', Philosophical Review, 20 [1982]. Rorty, A. O. (ed.) [1]. The Identities of Persons (Berkeley: University of California Press, 1976). Rosch, Eleanor, Mevis, C. et al. [1]. 'Basic Objects in Natural Categories', Cognitive Psychology, 8 (1976). Routley, Richard, and MacRae, Vera [1]. 'On the Identity of Sensations and Physiological Occurrences', American Philosophical Quarterly, 3 (1966). Russell, Bertrand (1]. The Principles of Mathematics (London: AlIen & Unwin; 2nd edn., 1937). - - [2]. 'The Philosophy of Logical Atomism', in Logic and Knowledge, ed. Robert C. Marsh (London: George Allen & Unwin, 1956). - - and Whitehead, Alfred [1]. Principia Mathematica (Cambridge: Cambridge University Press; 2nd edn., 1925/7). Sacks, H. [1]. The Man Who Mistook His Wife for a Hat (London: Duckworth, 1985). Salmon, N. U. [1]. Reference and Essence (Oxford: Blackwell, 1982). Sanford, David H. [1]. 'Locke, Leibniz and Wiggins on Being in the Same Place at the Same Time', Philosophical Review, 79 (1970). SchefHer, S [1]. 'Ethics, Personal Identity and Ideals of the Person', Canadian Journal of Philosophy, 12 (1982). Schlesinger, G. [1]. 'Spatial, Temporal and Cosmic Parts', Southern Journal of Philosophy, 23 (1985). - - [2]. The Intelligibility of Nature (Aberdeen: Aberdeen University Press, 1985). Schwartz, S. P. (ed.) [1]. Naming, Necessity and Natural Kinds (lthaca: Corn ell University Press, 1977).



Schwartz, Stephen P. (1]. 'Natural Kinds and Nominal Kinds', Mind, 89 (1980). Scott, Dana [1]. 'Existence and Description in Formal Logic', in Bertrand Russell: Philosopher of the Century, ed. Ralph Schoenman (London: George Allen & Unwin, 1967). Shoemaker, Sydney [1]. Self-Knowledge and Self-Identity (Ithaca: Comell University Press, 1963). - - [2]. 'Wiggins on Identity', Philosophical Review, 79 (1970). - - [3]. 'Persons and Their Pasts', American Philosophical Quarterly, 7 [1970]. - - [4]. 'Identity, Properties and Causality', Midwest Studies in Philosophy, 4 (1979). - - [5]. 'Personal Identity: a Materialist's Account', in Shoemaker and Swinbume [1]. - - [6]. Identity, Cause and Mind (Cambridge: Cambridge University Press, 1984). - - [7]. Critical Notice: Parfit, Reasons and Persons, Mind, 44 (1985). - - and Swinbume, R. G. [1]. Personal Identity (Oxford: Blackwell, 1984). Shorter, J. M. [1]. 'More About Bodily Continuity and Personal Identity', Analysis, 22 (1961/2). Simons, P. [1]. 'Token Resistance', Analysis, 42 (1982). Slote, Michael [1]. 'Causality and the Concept of a "Thing" " Midwest Studies in Philosophy, 4 (1979). Smart, B. [1]. 'How to Reidentify the Ship ofTheseus', Analysis, 32 (1972). - - [2]. 'The Ship of Theseus, the Parthenon, and Disassembled Objects', Analysis, 34 (1973). - - [3]. 'Diachronous and Synchronous Selves', Canadian Journal of Philosophy, 6 (1976). Smart, J. J. C. [1]. 'Spatialising Time', Mind, 64 (1955). Sober, E. [1]. 'Sets, Species and Evolution', Philosophy of Science, 51 (1984). Stalnaker, R. [1]. Inquiry (Cambridge, Mass.: MIT Press, 1984). Stevenson, Leslie [1]. 'Relative Identity and Leibniz's Law', Philosophical Quarterly, 22 (1972), 155-8. - - [2]. 'A Formal Theory of Sortal Quantification', Notre Dame Journal of Formal Logic, 16 (1975), 185-207. Strawson, P. F. [1]. Individuals (London: Methuen, 1959). - - [2]. 'Particular and General', Proceedings of the Aristotelian Society, 54 (1954),233-60. - - [3]. The Bounds of Sense (London: Methuen, 1966). - - [4]. 'Chisholm on Identity Through Time', in Kiefer, H. E. and Munitz, Milton K. (eds.), Language, Belief, and Metaphysics (Albany: State University of New York Press, 1970).



- - [5]. 'Entity and Identity', in Lewis, H. D. (ed.), British Contempor~ry Philosophy (London: Alien & Unwin, 1976), Vol. IV. - - [6]. 'May Bes and Might Have Beens', in Margalit, A. (ed.), Meaning and Use (Dorclrecht: Reidel, 1979). Stroll, Avrum [1]. 'Faces', Inquiry, 28 (1985). Swinburne,R. G. [1]. Space and Time (New York: StMartins Press, 1968). - - [2]. 'Personal Identity', Proceedings of the Aristotelian Society, suppl. vol. (1973/4). - - [3]. 'Personal Identity: the Dualist Theory', in Shoemaker and Swinburne [1]. Tarski, Alfred [1]. 'The Semantic Conception of Truth', in Logic, Semantics and Metamathematics (Oxford: Oxford University Press, 1956). Thomason, R. H. [1]. 'A Semantic Theory of Sortal Incorrectness', Journal of Philosophical Logic, 1 (1972). Thomson, J. J. [1]. 'Parthood and Identity Across Time', Journal of Philosophy, 80 (1983). Tichy, Pavel [1]. 'Kripke on Necessity A Posteriori', Philosophical Studies, 43 (1983). Travis, C. [1]. 'Vagueness, Observation and Sorites', Mind, 44 (1985). Tye, Michael [1]. 'The Puzzle of Hesperus and Phosphorus', Australasian Journal of Philosophy, 56 (1978). Unger, P. [1]. 'I Do Not Exist', in MacDonald [1]. Vision, Gerald [1]. 'Essentialism and the Senses of Proper Names', American Philosophical Quarterly, 7 (1970). Wallace, John R. [1]. 'Sortal Predicates and Quantification', Journal of Philosophy, 62 (1965). - - [2]. 'On the Frame of Reference', in Davidson, D. and Harman, G. (eds.), The Semantics of Natural Language (Dorclrecht: D. Reidel, 1971). White, Alan [1]. 'Mind-Brain Analogies', Canadian Journal of Philosophy, 1 (1972). White, Roger [1]. 'Wittgenstein on Identity', Proceedings of the Aristotelian Society, 68 (1978). Whiny, C. M. and Zangwill, O. L. (eds.) [1]. Amnesia, 2nd edn. (London: Butterworths, 1977). Wiggins, D. [1]. 'The Individuation of Things and Places', Proceedings of the Aristotelian Society, Suppl. Vol. 37 (1963). - - [2]. Identity and Spatia-Temporal Continuity (Oxford: Blackwell, 1967). - - [3]. 'Identity-Statements', in R. J. Butler (ed.), Analytical Philosophy, Second Series (Oxford: Blackwell, 1965). - - [4]. 'Reply to Mr Chandler', Analysis, 29 (1968/9). - - [5]. 'On Being in the Same Place at the Same Time', Philosophical Review, 77 (1968).



- - [6]. 'Essentialism, Continuity and Identity', Synthese, 23 (1974). - - [7]. 'Locke, Butler and the Stream of Consciousness', in Rorty [1]. - - [8]. 'The Concern to Survive', Midwest Studies in Philosophy, 4 (1979). - - [9]. Sameness and Substance, (Oxford: Basil Blackwell, 1980). WilIiams, B. A. O. [1]. 'Personal Identity and Individuation', Proceedings of the Aristotelian Society, 57 (1956-7). - - [2]. 'Bodily Continuity and Personal Identity', Analysis, 21 (1960). - - [3]. 'Persons, Character and Morality', in Rorty [1]. - - [4]. 'The Self and the Future', Philosophical Review, 79 (1970), and in WiIliams [5]. - - [5]. Problems of the Self (Cambridge: Cambridge University Press, 1973). Wilson, I. R. [1]. 'Explanatory and Inferential Conditionals', Philosophical Studies, 35 (1979). Wilson, lan [1]. Engineering Solids (London: McGraw HilI, 1979). Wilson, N. L. [1]. 'Space, Time and Individuals' ,Journal of Philosophy, 52 (1955). Witggenstein, Ludwig [1]. Tractatus Logico-Philosophicus, translated by D. F. Pears and B. F. McGuinness (London: Routledge & Kegan Paul, 1961). - - [2]. The Blue and Brown Books (Oxford: Blackwell, 1958). - - [3]. Philosophical Investigations, translated by G. E. M. Anscombe (Oxford: Blackwell, 1953). Woodger, J. H. [1]. The Axiomatic Method in Biology (Cambridge: Cambridge University Press, 1937). Woods, John [1]. 'Essentialism, Self-Identity and Quantifying In', in Munitz [1]. - - [2]. ' "Non in Particular" ',CanadianJournal of Philosophy, 2 (1973). - - [3]. 'Identity and Modality', Philosophia, 5 (1975). Woods, M. J. [1]. 'The Individuation of Things and Places', Proceedings of the Aristotelian Society, Suppl. Vol. 37 (1963). - - [2]. 'Identity and Individuation', in Butler, R. J. (ed.), Analytical Philosophy (Oxford: Blackwell, 1965); 2nd ser. Zemach, E. M. [1]. 'In Defence of Relative Identity', Philosophical Studies, 26 (1974), 207-18.


accessibility 161, 165-6 affordance 188-9 aggregates 53, 89-90 see also collections Alexander of Aphrodisias 171 n. Ames, L. 81 n. amnesia 268, 296-301, 303-7, 346-7 and branching 299-300 analysis conceptual 62-3 see also philosophy animal persons 242 Archers (example) 108-9 Aristotle 63 artefacts 31, 70--3, 161-2, 166-8, 180--90,228 artefact types 71-2, 166-7,202 repair vs. replacement 11, 178, 185 Aschkenasy, J. 213 n. atomism 170--1, 178 atoms 180 Ayer, A. J. 278 Beatrice (example) 78-S0, 317-19 Bennett, J. 270 n. Berkeley, G. 48 'best candidate' theory 121-2 see also 'only x and y' rule bodies 47-8, 94, 216, 305-6, 308-11, 312,335 see also particulars brains 276, 321-2, 334, 336-9, 341--4, 351 Broad, C. D. 242 n. Brody, B. 14 n. bundle theory 269-70, 276-9, 282--4, 287-9,284-5,287-9,290,293-6, 303--4 Butler, J. 116, 119,250 n. Cantor, G. 207-8 capacities see skills carbon 179-80

Camap, R. 209, 234-6 Cartwright, H. M. 195, 197 causal condition 26-9,54,99, 107-9, 152-3,302--4,318-19,323--4,343 cause 99-107, 151-2, 154-5, 168,291 chain theory 33-7, 56, 90,168,190, 350 change 11,24,44-5,47,49,54,64-5, 126-7,152-3,162,178-9,183-5, 190,193,197-8,293,304 Cambridge 24-5 see also persistence j continuity Chisholm R. 156,260,276-7,279, 282-5 Chisholm's paradox 156 n. city-block metric 223--4 Clark, A. J. 176 Clark, E. V. 222 n., 226 n. closest continuer theory see continuity co-consciousness 246-9, 254, 263 see also connectedness coherence 296-8, 301, 311, 313-14, 322-3 collections 36, 123--4, 147, 260--2 see also mereology colour 204-5 companionship problem 235-6 components 96, 126-8, 146, 149-50, 157-8, 162-3, 170, 172--4, 178, 181-3, 184-6, 341-3 composition 39, 95-6, 185 and essence 159-61 concepts 191 n. see also analysis j sortals j natural kinds conceptual development 58-9, 62--4, 196,243,292,339,345-8 concrete 34, 95-7 connectedness 248, 294-5, 297-8, 301, 304,306-7,313,316,322 consciousness 241, 332-3 constitution 145, 160--4, 165, 166, 167, 199,260--2 see also composition

Index continuant 12,86,88,96, 136, 193 continuity compositional 38-9 closest continuer theory 41-2, 115-16 Hollywood 41-2, 59, 94-5, 293 psychological 293-6, 297 spatio-temporall1-13, 37-45, 97-9, 105-6, 125-6, 136, 295, 335, 353 theory 12, 15,58-9, 137, 185, 193-4 see also connectedness; discontinuity Coolidge, C. 1 copying processes 75-9,112,168,174, 244-5,317,342-4 Corben, J. 204 n. correlational structure 188-9, 219, 227, 228-32,234-6 cosmic pans see pans counterpans 131-4, 158, 159 n., 162-4, 165-6 Craig, W. 106 criteria 87, 96, 245, 251-5, 262 critical realism 220 crystal structure 104, 136, 172, 179-80 Dawkins, R. 175, 244 n. death 31~17, 322 deferred ostension 195-6 Democritus 170-1 Dennett, D. C. 59,65,82, 181,334 Descanes, R. 258, 320 n. design 181-3,203 Deutscher, M. 250 n Dewey, J. 352 diachronic identity 87 diachronic relations 146-8,269-70, 277-9,297-300 differentiation 213, 219, 221-2, 225-7 dimensions 218-21, 22:>-5, 232 discontinuity 43-4,91-5,97-9, 136, 304,332-3 functional 92, 95-6 spatial 98, 113 disembodied existence 240-1 dualism 257-9, 320-1, 337 Duhem, P. 104, 347-8 Dummen, M. 164 n. Dumyat (example) 93-9, 107, 109-14, 127 Elliot, R. 301 n. episodic objects see discontinuity


epistemology 51, 146-8, 167, 176,205, 217,272 essence 150, 155, 159-61, 165-6, 170, 190-1,193,202 eternal sentences 135 Euclidean distance 223 Eusa problem 62-5, 85,126,147,197, 353 Evans, G. 279-82 events 151-2 existence contingent vs. necessary 163 n. see also metaphysics experience 51, 59-61, 140 explanation 101, 109-10, 144 D-N model 101-4 extension 209-11 extensionalist myth 207-11, 215-17, 228,232-4 extensionalism 164 n, 212-15, 229, 231 family resemblance 233 fifth dimension 133,261,354-5 see also pans, cosmic figure and ground 220-1 fission 254 n., 262-7,299, 304-5, 328-9 Forbes, G. 121 n., 123 n., 159 n., 163, 164 foundationalism 58 n., 285-6 Frege, G. 40 n., 209-10, 252 n. function 181-3, 184-7, 188-90 functional discontinuity 21, 38-9, 41, 92,95 see also pans functionalism 257-8 Funh, M. 209 n. fuzzy logic 163-4 Garner, W. R. 218-32 passim Geach, P. T. 192 n. genes 137, 168, 175,201,224 Ghiselin, M. T. 353 n. Gibbard, A. 52, 53 n., 88-90 Gibbens, T. C. N. 347 Gibson, J. J. 188-9,218 Goldstein, L. 77 n. Goodrnan, N. 35, 97, 102, 109 n., 149, 211 n., 212 n., 234 gravity 100, 1O~7 Griffin, N. 7 n., 10 n., 192 n., 193 n. growth 198 Guy Fawkes (example) 115-16,262-5



haecceitism 159-66 Hall Williams, J. 347 Hanson, N. R. 189 Hempel, C. G. 101-3 Heraclitus 5, 11 Hirsch, E. 7 n., 38, 43-4, 47--8, 52, 55 n., 91, 94-6,115,119,180, 214,293-4,342 histories 23, 54, 110, 145, 149, 160, 199 Hoban, R. 63 Hobbes, T. 11 Hofstadter, D. R. 65 Hull, D. 353 n. Hume, D. 23, 25, 48, 85, 91, 99-100, 103-7,137--8,154,207,212, 214-15,261-2,268-70 identification-free judgements 279--81 identity 136, 191-9,260-1,266,282-4 conditions of 5-32 passim and importance 122 of indiscernibles 10 necessity of 116, 119, 155 primitiveness of 40,51--8 qualitative see similarity relative 192-3 reflexivity of 9 see also personal identity; similarity; survival ideology 136 indeterminacy 9, 22-3, 290, 347--8 indiscemibility principle of 9 theory 11, 13-17 individuals 34-6, 98-9, 118-24, 149, 157--8 infatuation 310-11 information 218-19 innateness 6, 48, 221 integrality see dimensions intension 209-10 Ishiguro, H. 10 n. Iversen, S. D. 296 l